{ "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": 0.11337890625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1526.8, "completions/mean_length": 487.7326171875, "completions/mean_terminated_length": 353.7248291015625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.0035850778222084045, "learning_rate": 3.1249999999999997e-07, "loss": 0.0442, "num_tokens": 19980846.0, "reward": 0.20237886905670166, "reward_std": 0.3675601720809937, "rewards/accuracy_reward": 0.05322265625, "rewards/brier_reward": 0.09767542034387589, "rewards/confidence_one_or_zero": 0.41123046875, "rewards/format_reward": 0.25390625, "rewards/mean_confidence_reward": 0.5377732753753662, "rewards/smce": -0.484735369682312, "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": 0.1056640625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1527.4, "completions/mean_length": 448.823046875, "completions/mean_terminated_length": 320.8543334960938, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.004688555840402842, "learning_rate": 6.249999999999999e-07, "loss": 0.045, "num_tokens": 39560218.0, "reward": 0.2146211951971054, "reward_std": 0.37262771725654603, "rewards/accuracy_reward": 0.05390625, "rewards/brier_reward": 0.10282630771398545, "rewards/confidence_one_or_zero": 0.435546875, "rewards/format_reward": 0.27255859375, "rewards/mean_confidence_reward": 0.5611979961395264, "rewards/smce": -0.5075305879116059, "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": 0.0509765625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1487.4, "completions/mean_length": 311.6556640625, "completions/mean_terminated_length": 246.38412780761718, "completions/min_length": 1.4, "completions/min_terminated_length": 1.4, "epoch": 0.048, "grad_norm": 0.004514631815254688, "learning_rate": 9.374999999999999e-07, "loss": 0.0503, "num_tokens": 57817620.0, "reward": 0.40357725620269774, "reward_std": 0.44316648244857787, "rewards/accuracy_reward": 0.10810546875, "rewards/brier_reward": 0.199105966091156, "rewards/confidence_one_or_zero": 0.5349609375, "rewards/format_reward": 0.5, "rewards/mean_confidence_reward": 0.7012235999107361, "rewards/smce": -0.5937792539596558, "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": 0.0224609375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1454.8, "completions/mean_length": 211.46875, "completions/mean_terminated_length": 181.1057159423828, "completions/min_length": 15.4, "completions/min_terminated_length": 15.4, "epoch": 0.064, "grad_norm": 0.0024940520524978638, "learning_rate": 1e-06, "loss": 0.0369, "num_tokens": 75034132.0, "reward": 0.6788389444351196, "reward_std": 0.3938281416893005, "rewards/accuracy_reward": 0.189453125, "rewards/brier_reward": 0.3284399390220642, "rewards/confidence_one_or_zero": 0.5779296875, "rewards/format_reward": 0.83984375, "rewards/mean_confidence_reward": 0.8039127707481384, "rewards/smce": -0.6165119409561157, "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": 0.01513671875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1392.2, "completions/mean_length": 163.4111328125, "completions/mean_terminated_length": 142.32491455078124, "completions/min_length": 20.8, "completions/min_terminated_length": 20.8, "epoch": 0.08, "grad_norm": 0.0027408404275774956, "learning_rate": 1e-06, "loss": 0.0234, "num_tokens": 91639238.0, "reward": 0.8109780430793763, "reward_std": 0.3374348938465118, "rewards/accuracy_reward": 0.2431640625, "rewards/brier_reward": 0.42249656319618223, "rewards/confidence_one_or_zero": 0.50439453125, "rewards/format_reward": 0.95634765625, "rewards/mean_confidence_reward": 0.7859874844551087, "rewards/smce": -0.547917366027832, "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": 0.0091796875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1320.8, "completions/mean_length": 145.26171875, "completions/mean_terminated_length": 132.38874359130858, "completions/min_length": 13.8, "completions/min_terminated_length": 13.8, "epoch": 0.096, "grad_norm": 0.0023242132738232613, "learning_rate": 1e-06, "loss": 0.0154, "num_tokens": 107931710.0, "reward": 0.8883284211158753, "reward_std": 0.2937169373035431, "rewards/accuracy_reward": 0.2525390625, "rewards/brier_reward": 0.5444689631462097, "rewards/confidence_one_or_zero": 0.3251953125, "rewards/format_reward": 0.9796875, "rewards/mean_confidence_reward": 0.6470721125602722, "rewards/smce": -0.4061504006385803, "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": 0.0044921875, "completions/max_length": 1536.0, "completions/max_terminated_length": 866.8, "completions/mean_length": 127.86181640625, "completions/mean_terminated_length": 121.51152038574219, "completions/min_length": 36.6, "completions/min_terminated_length": 36.6, "epoch": 0.112, "grad_norm": 0.0018074740655720234, "learning_rate": 1e-06, "loss": 0.0098, "num_tokens": 124109815.0, "reward": 0.9837763190269471, "reward_std": 0.2189751923084259, "rewards/accuracy_reward": 0.2767578125, "rewards/brier_reward": 0.69852614402771, "rewards/confidence_one_or_zero": 0.18828125, "rewards/format_reward": 0.99228515625, "rewards/mean_confidence_reward": 0.4274466037750244, "rewards/smce": -0.17678792178630828, "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": 0.00244140625, "completions/max_length": 1536.0, "completions/max_terminated_length": 691.0, "completions/mean_length": 124.60478515625, "completions/mean_terminated_length": 121.14984130859375, "completions/min_length": 35.4, "completions/min_terminated_length": 35.4, "epoch": 0.128, "grad_norm": 0.0009728963486850262, "learning_rate": 1e-06, "loss": 0.0049, "num_tokens": 140367656.0, "reward": 1.0210827350616456, "reward_std": 0.15196589678525924, "rewards/accuracy_reward": 0.28193359375, "rewards/brier_reward": 0.7644367456436157, "rewards/confidence_one_or_zero": 0.146875, "rewards/format_reward": 0.99580078125, "rewards/mean_confidence_reward": 0.28299830853939056, "rewards/smce": -0.06476258486509323, "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": 0.0005859375, "completions/max_length": 1318.4, "completions/max_terminated_length": 633.8, "completions/mean_length": 121.4595703125, "completions/mean_terminated_length": 120.63011322021484, "completions/min_length": 48.4, "completions/min_terminated_length": 48.4, "epoch": 0.144, "grad_norm": 0.0008337603649124503, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 156501322.0, "reward": 1.0360254526138306, "reward_std": 0.11216316372156143, "rewards/accuracy_reward": 0.3078125, "rewards/brier_reward": 0.7661016464233399, "rewards/confidence_one_or_zero": 0.10185546875, "rewards/format_reward": 0.99814453125, "rewards/mean_confidence_reward": 0.23704092800617219, "rewards/smce": -0.08523178026080132, "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": 0.00068359375, "completions/max_length": 1308.0, "completions/max_terminated_length": 469.2, "completions/mean_length": 122.78017578125, "completions/mean_terminated_length": 121.81385955810546, "completions/min_length": 54.0, "completions/min_terminated_length": 54.0, "epoch": 0.16, "grad_norm": 0.0008199186413548887, "learning_rate": 1e-06, "loss": 0.0016, "num_tokens": 172552831.0, "reward": 1.0488179683685304, "reward_std": 0.1010824665427208, "rewards/accuracy_reward": 0.3365234375, "rewards/brier_reward": 0.7621974587440491, "rewards/confidence_one_or_zero": 0.06181640625, "rewards/format_reward": 0.99892578125, "rewards/mean_confidence_reward": 0.22926148474216462, "rewards/smce": -0.11216033399105071, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 401.0, "eval_completions/max_terminated_length": 401.0, "eval_completions/mean_length": 125.22658920288086, "eval_completions/mean_terminated_length": 125.22658920288086, "eval_completions/min_length": 63.0, "eval_completions/min_terminated_length": 63.0, "eval_loss": 0.0, "eval_num_tokens": 172552831.0, "eval_reward": 1.032656490802765, "eval_reward_std": 0.1260393038392067, "eval_rewards/accuracy_reward": 0.271484375, "eval_rewards/brier_reward": 0.7938330173492432, "eval_rewards/confidence_one_or_zero": 0.041015625, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.23232988268136978, "eval_rewards/smce": -0.04869849234819412, "eval_runtime": 12.3556, "eval_samples_per_second": 40.468, "eval_steps_per_second": 0.162, "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": 0.0005859375, "completions/max_length": 1308.6, "completions/max_terminated_length": 431.8, "completions/mean_length": 125.6203125, "completions/mean_terminated_length": 124.79365539550781, "completions/min_length": 56.8, "completions/min_terminated_length": 56.8, "epoch": 0.176, "grad_norm": 0.0008834801847115159, "learning_rate": 1e-06, "loss": 0.0016, "num_tokens": 188904367.0, "reward": 1.0448211193084718, "reward_std": 0.09679851233959198, "rewards/accuracy_reward": 0.31142578125, "rewards/brier_reward": 0.779005253314972, "rewards/confidence_one_or_zero": 0.03623046875, "rewards/format_reward": 0.99921875, "rewards/mean_confidence_reward": 0.24507814943790435, "rewards/smce": -0.08033295050263405, "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": 0.000390625, "completions/max_length": 1106.6, "completions/max_terminated_length": 387.2, "completions/mean_length": 124.562109375, "completions/mean_terminated_length": 124.01052703857422, "completions/min_length": 56.4, "completions/min_terminated_length": 56.4, "epoch": 0.192, "grad_norm": 0.0007722419104538858, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 205241867.0, "reward": 1.050026798248291, "reward_std": 0.09879964888095856, "rewards/accuracy_reward": 0.3302734375, "rewards/brier_reward": 0.77076256275177, "rewards/confidence_one_or_zero": 0.0201171875, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.2705323278903961, "rewards/smce": -0.06530658006668091, "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": 0.00029296875, "completions/max_length": 822.2, "completions/max_terminated_length": 369.6, "completions/mean_length": 125.0603515625, "completions/mean_terminated_length": 124.64741363525391, "completions/min_length": 54.6, "completions/min_terminated_length": 54.6, "epoch": 0.208, "grad_norm": 0.0008552237995900214, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 221543093.0, "reward": 1.060682201385498, "reward_std": 0.10427201688289642, "rewards/accuracy_reward": 0.34794921875, "rewards/brier_reward": 0.7738091349601746, "rewards/confidence_one_or_zero": 0.00810546875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3167231798171997, "rewards/smce": -0.038938027992844584, "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": 0.0, "completions/max_length": 366.4, "completions/max_terminated_length": 366.4, "completions/mean_length": 123.75830078125, "completions/mean_terminated_length": 123.75830078125, "completions/min_length": 56.2, "completions/min_terminated_length": 56.2, "epoch": 0.224, "grad_norm": 0.0009770386386662722, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 237879594.0, "reward": 1.0563500642776489, "reward_std": 0.10786637514829636, "rewards/accuracy_reward": 0.32978515625, "rewards/brier_reward": 0.7832106351852417, "rewards/confidence_one_or_zero": 0.00361328125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.34383692145347594, "rewards/smce": -0.033423048071563245, "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": 9.765625e-05, "completions/max_length": 550.0, "completions/max_terminated_length": 336.2, "completions/mean_length": 123.43134765625, "completions/mean_terminated_length": 123.29346313476563, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.24, "grad_norm": 0.0008802193333394825, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 254230027.0, "reward": 1.061513566970825, "reward_std": 0.10092030316591263, "rewards/accuracy_reward": 0.3501953125, "rewards/brier_reward": 0.7731273651123047, "rewards/confidence_one_or_zero": 0.0021484375, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.34638115763664246, "rewards/smce": -0.03192154616117478, "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": 9.765625e-05, "completions/max_length": 596.6, "completions/max_terminated_length": 356.6, "completions/mean_length": 123.50390625, "completions/mean_terminated_length": 123.36609649658203, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.256, "grad_norm": 0.0008312573190778494, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 270830611.0, "reward": 1.0700663805007935, "reward_std": 0.10508463829755783, "rewards/accuracy_reward": 0.35947265625, "rewards/brier_reward": 0.7807585239410401, "rewards/confidence_one_or_zero": 0.00205078125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.35435292720794676, "rewards/smce": -0.013978258706629276, "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": 0.00029296875, "completions/max_length": 1058.2, "completions/max_terminated_length": 351.2, "completions/mean_length": 123.29365234375, "completions/mean_terminated_length": 122.8795181274414, "completions/min_length": 61.8, "completions/min_terminated_length": 61.8, "epoch": 0.272, "grad_norm": 0.0009743980481289327, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 287208562.0, "reward": 1.078542709350586, "reward_std": 0.11005425304174424, "rewards/accuracy_reward": 0.39365234375, "rewards/brier_reward": 0.7639232754707337, "rewards/confidence_one_or_zero": 0.00087890625, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.3724678099155426, "rewards/smce": -0.026533697731792927, "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": 372.4, "completions/max_terminated_length": 372.4, "completions/mean_length": 122.5517578125, "completions/mean_terminated_length": 122.5517578125, "completions/min_length": 57.4, "completions/min_terminated_length": 57.4, "epoch": 0.288, "grad_norm": 0.0009167360258288682, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 303624452.0, "reward": 1.0746608257293702, "reward_std": 0.10677139461040497, "rewards/accuracy_reward": 0.37080078125, "rewards/brier_reward": 0.7785232663154602, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.38262552618980405, "rewards/smce": -0.031942544505000114, "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": 0.0, "completions/max_length": 326.6, "completions/max_terminated_length": 326.6, "completions/mean_length": 121.0849609375, "completions/mean_terminated_length": 121.0849609375, "completions/min_length": 59.6, "completions/min_terminated_length": 59.6, "epoch": 0.304, "grad_norm": 0.0009152097627520561, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 319829130.0, "reward": 1.0819915771484374, "reward_std": 0.10794731378555297, "rewards/accuracy_reward": 0.40126953125, "rewards/brier_reward": 0.7629124164581299, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4008802831172943, "rewards/smce": -0.04182044714689255, "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": 0.0, "completions/max_length": 280.6, "completions/max_terminated_length": 280.6, "completions/mean_length": 117.4109375, "completions/mean_terminated_length": 117.4109375, "completions/min_length": 60.4, "completions/min_terminated_length": 60.4, "epoch": 0.32, "grad_norm": 0.0009640919743105769, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 335988634.0, "reward": 1.0821470260620116, "reward_std": 0.1047831878066063, "rewards/accuracy_reward": 0.39453125, "rewards/brier_reward": 0.7697669267654419, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4091184735298157, "rewards/smce": -0.049343763291835784, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 202.5, "eval_completions/max_terminated_length": 202.5, "eval_completions/mean_length": 114.07200622558594, "eval_completions/mean_terminated_length": 114.07200622558594, "eval_completions/min_length": 60.0, "eval_completions/min_terminated_length": 60.0, "eval_loss": 0.0, "eval_num_tokens": 335988634.0, "eval_reward": 1.0488357543945312, "eval_reward_std": 0.1998460590839386, "eval_rewards/accuracy_reward": 0.314453125, "eval_rewards/brier_reward": 0.783225953578949, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.3979492336511612, "eval_rewards/smce": -0.0844060592353344, "eval_runtime": 9.2582, "eval_samples_per_second": 54.006, "eval_steps_per_second": 0.216, "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": 0.0, "completions/max_length": 314.0, "completions/max_terminated_length": 314.0, "completions/mean_length": 116.25380859375, "completions/mean_terminated_length": 116.25380859375, "completions/min_length": 59.0, "completions/min_terminated_length": 59.0, "epoch": 0.336, "grad_norm": 0.0010559717193245888, "learning_rate": 1e-06, "loss": -0.0004, "num_tokens": 352170305.0, "reward": 1.0908453702926635, "reward_std": 0.10500690937042237, "rewards/accuracy_reward": 0.40830078125, "rewards/brier_reward": 0.7734902620315551, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.3929440498352051, "rewards/smce": -0.03384199421852827, "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": 0.0, "completions/max_length": 366.8, "completions/max_terminated_length": 366.8, "completions/mean_length": 121.5208984375, "completions/mean_terminated_length": 121.5208984375, "completions/min_length": 62.0, "completions/min_terminated_length": 62.0, "epoch": 0.352, "grad_norm": 0.0010314425453543663, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 368524887.0, "reward": 1.0825308561325073, "reward_std": 0.10590031743049622, "rewards/accuracy_reward": 0.392578125, "rewards/brier_reward": 0.7725841403007507, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.3955056965351105, "rewards/smce": -0.036800354905426505, "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": 341.2, "completions/max_terminated_length": 341.2, "completions/mean_length": 133.803515625, "completions/mean_terminated_length": 133.803515625, "completions/min_length": 57.8, "completions/min_terminated_length": 57.8, "epoch": 0.368, "grad_norm": 0.0009640640346333385, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 384988507.0, "reward": 1.0816544771194458, "reward_std": 0.10436138659715652, "rewards/accuracy_reward": 0.3865234375, "rewards/brier_reward": 0.7767881631851197, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4014230787754059, "rewards/smce": -0.034000173956155774, "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": 9.765625e-05, "completions/max_length": 702.4, "completions/max_terminated_length": 454.8, "completions/mean_length": 146.3341796875, "completions/mean_terminated_length": 146.19812927246093, "completions/min_length": 66.4, "completions/min_terminated_length": 66.4, "epoch": 0.384, "grad_norm": 0.0008643779437988997, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 401478553.0, "reward": 1.0923637866973877, "reward_std": 0.10116445124149323, "rewards/accuracy_reward": 0.4091796875, "rewards/brier_reward": 0.7758468389511108, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3995400667190552, "rewards/smce": -0.06742004379630089, "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": 613.6, "completions/max_terminated_length": 379.0, "completions/mean_length": 149.774609375, "completions/mean_terminated_length": 149.63934326171875, "completions/min_length": 71.6, "completions/min_terminated_length": 71.6, "epoch": 0.4, "grad_norm": 0.0008598100976087153, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 417872245.0, "reward": 1.0924156427383422, "reward_std": 0.11007419526576996, "rewards/accuracy_reward": 0.4076171875, "rewards/brier_reward": 0.77731374502182, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.404058700799942, "rewards/smce": -0.028200492449104785, "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": 665.2, "completions/max_terminated_length": 443.2, "completions/mean_length": 156.9810546875, "completions/mean_terminated_length": 156.57687683105468, "completions/min_length": 73.4, "completions/min_terminated_length": 73.4, "epoch": 0.416, "grad_norm": 0.0010003651259467006, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 434623475.0, "reward": 1.110920262336731, "reward_std": 0.11143997013568878, "rewards/accuracy_reward": 0.45322265625, "rewards/brier_reward": 0.7691106915473938, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.41947434544563295, "rewards/smce": -0.05290180444717407, "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": 393.6, "completions/max_terminated_length": 393.6, "completions/mean_length": 163.24091796875, "completions/mean_terminated_length": 163.24091796875, "completions/min_length": 72.2, "completions/min_terminated_length": 72.2, "epoch": 0.432, "grad_norm": 0.0007602354162372649, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 451437494.0, "reward": 1.0949999332427978, "reward_std": 0.11146767735481262, "rewards/accuracy_reward": 0.41318359375, "rewards/brier_reward": 0.7768206119537353, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4301083981990814, "rewards/smce": -0.052707664296031, "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": 0.0, "completions/max_length": 427.0, "completions/max_terminated_length": 427.0, "completions/mean_length": 165.24130859375, "completions/mean_terminated_length": 165.24130859375, "completions/min_length": 76.2, "completions/min_terminated_length": 76.2, "epoch": 0.448, "grad_norm": 0.000916799355763942, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 468191741.0, "reward": 1.1065703868865966, "reward_std": 0.11395464688539506, "rewards/accuracy_reward": 0.4322265625, "rewards/brier_reward": 0.7810142278671265, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.42705950140953064, "rewards/smce": -0.032019305787980554, "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": 0.0, "completions/max_length": 397.8, "completions/max_terminated_length": 397.8, "completions/mean_length": 169.6619140625, "completions/mean_terminated_length": 169.6619140625, "completions/min_length": 79.4, "completions/min_terminated_length": 79.4, "epoch": 0.464, "grad_norm": 0.0007820753962732852, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 484876247.0, "reward": 1.109752631187439, "reward_std": 0.10953254997730255, "rewards/accuracy_reward": 0.44111328125, "rewards/brier_reward": 0.7784909725189209, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4319294333457947, "rewards/smce": -0.021954568102955818, "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": 947.6, "completions/max_terminated_length": 490.4, "completions/mean_length": 171.62763671875, "completions/mean_terminated_length": 171.36109313964843, "completions/min_length": 81.2, "completions/min_terminated_length": 81.2, "epoch": 0.48, "grad_norm": 0.0009613652364350855, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 501703954.0, "reward": 1.1193360567092896, "reward_std": 0.1150269329547882, "rewards/accuracy_reward": 0.46259765625, "rewards/brier_reward": 0.7764689922332764, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.43756228089332583, "rewards/smce": -0.047635878622531894, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 353.0, "eval_completions/max_terminated_length": 353.0, "eval_completions/mean_length": 173.3000717163086, "eval_completions/mean_terminated_length": 173.3000717163086, "eval_completions/min_length": 96.5, "eval_completions/min_terminated_length": 96.5, "eval_loss": 0.0, "eval_num_tokens": 501703954.0, "eval_reward": 1.0731765031814575, "eval_reward_std": 0.22983557730913162, "eval_rewards/accuracy_reward": 0.36328125, "eval_rewards/brier_reward": 0.7830783128738403, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.43638673424720764, "eval_rewards/smce": -0.07412109896540642, "eval_runtime": 12.0771, "eval_samples_per_second": 41.401, "eval_steps_per_second": 0.166, "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": 9.765625e-05, "completions/max_length": 605.8, "completions/max_terminated_length": 368.0, "completions/mean_length": 173.3896484375, "completions/mean_terminated_length": 173.25641174316405, "completions/min_length": 87.0, "completions/min_terminated_length": 87.0, "epoch": 0.496, "grad_norm": 0.0009212964214384556, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 518601576.0, "reward": 1.118337345123291, "reward_std": 0.110992231965065, "rewards/accuracy_reward": 0.4560546875, "rewards/brier_reward": 0.780720853805542, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44267138838768005, "rewards/smce": -0.04033624045550823, "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": 0.0, "completions/max_length": 418.8, "completions/max_terminated_length": 418.8, "completions/mean_length": 174.5041015625, "completions/mean_terminated_length": 174.5041015625, "completions/min_length": 86.6, "completions/min_terminated_length": 86.6, "epoch": 0.512, "grad_norm": 0.0007764216279610991, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 535378450.0, "reward": 1.1107932567596435, "reward_std": 0.10566790401935577, "rewards/accuracy_reward": 0.4388671875, "rewards/brier_reward": 0.7827236652374268, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.43680567145347593, "rewards/smce": -0.051772523671388626, "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": 634.8, "completions/max_terminated_length": 437.2, "completions/mean_length": 172.83876953125, "completions/mean_terminated_length": 172.7058319091797, "completions/min_length": 83.2, "completions/min_terminated_length": 83.2, "epoch": 0.528, "grad_norm": 0.0008587085176259279, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 552035807.0, "reward": 1.1129685878753661, "reward_std": 0.10902059525251388, "rewards/accuracy_reward": 0.4474609375, "rewards/brier_reward": 0.7785768270492553, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.45129268169403075, "rewards/smce": -0.038398644607514146, "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": 0.0001953125, "completions/max_length": 678.0, "completions/max_terminated_length": 462.6, "completions/mean_length": 171.39228515625, "completions/mean_terminated_length": 171.1254852294922, "completions/min_length": 84.0, "completions/min_terminated_length": 84.0, "epoch": 0.544, "grad_norm": 0.0009459138382226229, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 568991088.0, "reward": 1.0985527276992797, "reward_std": 0.10920530259609222, "rewards/accuracy_reward": 0.41337890625, "rewards/brier_reward": 0.7839253544807434, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4351104199886322, "rewards/smce": -0.04374492410570383, "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": 641.0, "completions/max_terminated_length": 424.8, "completions/mean_length": 174.48369140625, "completions/mean_terminated_length": 174.3507507324219, "completions/min_length": 84.4, "completions/min_terminated_length": 84.4, "epoch": 0.56, "grad_norm": 0.0009580635232850909, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 585903977.0, "reward": 1.1115816354751586, "reward_std": 0.10534543991088867, "rewards/accuracy_reward": 0.43564453125, "rewards/brier_reward": 0.787619686126709, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4416061580181122, "rewards/smce": -0.04100565351545811, "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": 396.6, "completions/max_terminated_length": 396.6, "completions/mean_length": 173.34697265625, "completions/mean_terminated_length": 173.34697265625, "completions/min_length": 81.8, "completions/min_terminated_length": 81.8, "epoch": 0.576, "grad_norm": 0.0008675847202539444, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 602717066.0, "reward": 1.1025778532028199, "reward_std": 0.09847408682107925, "rewards/accuracy_reward": 0.4197265625, "rewards/brier_reward": 0.7854331374168396, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4456318259239197, "rewards/smce": -0.04764346107840538, "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": 0.00029296875, "completions/max_length": 712.8, "completions/max_terminated_length": 498.4, "completions/mean_length": 173.9390625, "completions/mean_terminated_length": 173.53940124511718, "completions/min_length": 84.8, "completions/min_terminated_length": 84.8, "epoch": 0.592, "grad_norm": 0.0009322913829237223, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 619666010.0, "reward": 1.12094247341156, "reward_std": 0.10624455362558365, "rewards/accuracy_reward": 0.4673828125, "rewards/brier_reward": 0.774797546863556, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4424082159996033, "rewards/smce": -0.03360339626669884, "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": 9.765625e-05, "completions/max_length": 625.2, "completions/max_terminated_length": 395.6, "completions/mean_length": 176.16884765625, "completions/mean_terminated_length": 176.03597717285157, "completions/min_length": 85.4, "completions/min_terminated_length": 85.4, "epoch": 0.608, "grad_norm": 0.0008757953182794154, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 636325915.0, "reward": 1.1331445455551148, "reward_std": 0.10666034519672393, "rewards/accuracy_reward": 0.47880859375, "rewards/brier_reward": 0.7875788450241089, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47118945717811583, "rewards/smce": -0.016280164942145348, "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": 0.0, "completions/max_length": 463.2, "completions/max_terminated_length": 463.2, "completions/mean_length": 179.38330078125, "completions/mean_terminated_length": 179.38330078125, "completions/min_length": 84.8, "completions/min_terminated_length": 84.8, "epoch": 0.624, "grad_norm": 0.0009400384151376784, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 653288912.0, "reward": 1.1178892612457276, "reward_std": 0.11412412077188491, "rewards/accuracy_reward": 0.44765625, "rewards/brier_reward": 0.7881268620491028, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.47951045632362366, "rewards/smce": -0.05635717511177063, "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": 9.765625e-05, "completions/max_length": 719.0, "completions/max_terminated_length": 520.8, "completions/mean_length": 180.0751953125, "completions/mean_terminated_length": 179.94290771484376, "completions/min_length": 85.6, "completions/min_terminated_length": 85.6, "epoch": 0.64, "grad_norm": 0.0008365341345779598, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 670308850.0, "reward": 1.0954386472702027, "reward_std": 0.11347412168979645, "rewards/accuracy_reward": 0.40654296875, "rewards/brier_reward": 0.7844377636909485, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4732663094997406, "rewards/smce": -0.06727753281593322, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 349.0, "eval_completions/max_terminated_length": 349.0, "eval_completions/mean_length": 184.80519104003906, "eval_completions/mean_terminated_length": 184.80519104003906, "eval_completions/min_length": 101.0, "eval_completions/min_terminated_length": 101.0, "eval_loss": 0.0, "eval_num_tokens": 670308850.0, "eval_reward": 1.0808968544006348, "eval_reward_std": 0.2522861063480377, "eval_rewards/accuracy_reward": 0.36328125, "eval_rewards/brier_reward": 0.7985226511955261, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.47593750059604645, "eval_rewards/smce": -0.11298829317092896, "eval_runtime": 12.1311, "eval_samples_per_second": 41.217, "eval_steps_per_second": 0.165, "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": 417.6, "completions/max_terminated_length": 417.6, "completions/mean_length": 180.13681640625, "completions/mean_terminated_length": 180.13681640625, "completions/min_length": 83.8, "completions/min_terminated_length": 83.8, "epoch": 0.656, "grad_norm": 0.0008805691613815725, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 687058955.0, "reward": 1.1146647214889527, "reward_std": 0.10650343894958496, "rewards/accuracy_reward": 0.4419921875, "rewards/brier_reward": 0.7873411774635315, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.48468555212020875, "rewards/smce": -0.04835297502577305, "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": 0.00029296875, "completions/max_length": 861.0, "completions/max_terminated_length": 467.2, "completions/mean_length": 182.21337890625, "completions/mean_terminated_length": 181.81690979003906, "completions/min_length": 89.6, "completions/min_terminated_length": 89.6, "epoch": 0.672, "grad_norm": 0.0007954036700539291, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 704059892.0, "reward": 1.1312153816223145, "reward_std": 0.10522501021623612, "rewards/accuracy_reward": 0.46982421875, "rewards/brier_reward": 0.7929009675979615, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.46643847823143003, "rewards/smce": -0.022859130799770356, "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": 0.0, "completions/max_length": 416.8, "completions/max_terminated_length": 416.8, "completions/mean_length": 186.283203125, "completions/mean_terminated_length": 186.283203125, "completions/min_length": 91.2, "completions/min_terminated_length": 91.2, "epoch": 0.688, "grad_norm": 0.000893253309186548, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 720923496.0, "reward": 1.130001473426819, "reward_std": 0.09947767555713653, "rewards/accuracy_reward": 0.46083984375, "rewards/brier_reward": 0.7991663455963135, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4343766748905182, "rewards/smce": -0.04040838629007339, "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": 746.2, "completions/max_terminated_length": 589.8, "completions/mean_length": 187.50205078125, "completions/mean_terminated_length": 187.37029724121095, "completions/min_length": 87.2, "completions/min_terminated_length": 87.2, "epoch": 0.704, "grad_norm": 0.000925455882679671, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 737821181.0, "reward": 1.1401437997817994, "reward_std": 0.10884704440832138, "rewards/accuracy_reward": 0.49033203125, "rewards/brier_reward": 0.7900588274002075, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47714354991912844, "rewards/smce": -0.06580102406442165, "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": 0.00029296875, "completions/max_length": 895.4, "completions/max_terminated_length": 455.0, "completions/mean_length": 187.88642578125, "completions/mean_terminated_length": 187.4909240722656, "completions/min_length": 92.2, "completions/min_terminated_length": 92.2, "epoch": 0.72, "grad_norm": 0.0009635393507778645, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 754833394.0, "reward": 1.1586825370788574, "reward_std": 0.10534119158983231, "rewards/accuracy_reward": 0.51337890625, "rewards/brier_reward": 0.8042813301086426, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.48370898962020875, "rewards/smce": -0.031232983246445654, "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": 0.0, "completions/max_length": 469.0, "completions/max_terminated_length": 469.0, "completions/mean_length": 186.92626953125, "completions/mean_terminated_length": 186.92626953125, "completions/min_length": 93.2, "completions/min_terminated_length": 93.2, "epoch": 0.736, "grad_norm": 0.0009669368155300617, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 771878079.0, "reward": 1.1391812324523927, "reward_std": 0.11643845587968826, "rewards/accuracy_reward": 0.48623046875, "rewards/brier_reward": 0.7922345280647278, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5351982235908508, "rewards/smce": -0.05878893509507179, "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": 751.2, "completions/max_terminated_length": 566.0, "completions/mean_length": 183.4779296875, "completions/mean_terminated_length": 183.34601135253905, "completions/min_length": 85.8, "completions/min_terminated_length": 85.8, "epoch": 0.752, "grad_norm": 0.0010671571362763643, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 788732317.0, "reward": 1.1412156581878663, "reward_std": 0.12025659680366516, "rewards/accuracy_reward": 0.4857421875, "rewards/brier_reward": 0.7967909932136535, "rewards/confidence_one_or_zero": 0.00087890625, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5363173723220825, "rewards/smce": -0.0527303846552968, "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": 9.765625e-05, "completions/max_length": 710.0, "completions/max_terminated_length": 519.6, "completions/mean_length": 183.93583984375, "completions/mean_terminated_length": 183.8036376953125, "completions/min_length": 88.4, "completions/min_terminated_length": 88.4, "epoch": 0.768, "grad_norm": 0.0010159483645111322, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 805502060.0, "reward": 1.1255807876586914, "reward_std": 0.11585851460695266, "rewards/accuracy_reward": 0.4578125, "rewards/brier_reward": 0.7936447501182556, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4842294991016388, "rewards/smce": -0.03694276623427868, "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": 686.8, "completions/max_terminated_length": 517.8, "completions/mean_length": 184.45654296875, "completions/mean_terminated_length": 184.32433166503907, "completions/min_length": 90.4, "completions/min_terminated_length": 90.4, "epoch": 0.784, "grad_norm": 0.0009062938624992967, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 822614735.0, "reward": 1.145763921737671, "reward_std": 0.1053706169128418, "rewards/accuracy_reward": 0.503125, "rewards/brier_reward": 0.7886038422584534, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4539990246295929, "rewards/smce": -0.06559372935444116, "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": 0.0, "completions/max_length": 458.4, "completions/max_terminated_length": 458.4, "completions/mean_length": 182.2365234375, "completions/mean_terminated_length": 182.2365234375, "completions/min_length": 89.6, "completions/min_terminated_length": 89.6, "epoch": 0.8, "grad_norm": 0.0009169210097752512, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 839356229.0, "reward": 1.1485914945602418, "reward_std": 0.11149813830852509, "rewards/accuracy_reward": 0.4984375, "rewards/brier_reward": 0.7987475156784057, "rewards/confidence_one_or_zero": 0.0015625, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5062128841876984, "rewards/smce": -0.030764539353549482, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 378.0, "eval_completions/max_terminated_length": 378.0, "eval_completions/mean_length": 186.12918853759766, "eval_completions/mean_terminated_length": 186.12918853759766, "eval_completions/min_length": 100.0, "eval_completions/min_terminated_length": 100.0, "eval_loss": 0.0, "eval_num_tokens": 839356229.0, "eval_reward": 1.097034513950348, "eval_reward_std": 0.2692975252866745, "eval_rewards/accuracy_reward": 0.396484375, "eval_rewards/brier_reward": 0.7975927591323853, "eval_rewards/confidence_one_or_zero": 0.001953125, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4864257723093033, "eval_rewards/smce": -0.09077637270092964, "eval_runtime": 12.4371, "eval_samples_per_second": 40.202, "eval_steps_per_second": 0.161, "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": 0.0, "completions/max_length": 489.8, "completions/max_terminated_length": 489.8, "completions/mean_length": 186.72607421875, "completions/mean_terminated_length": 186.72607421875, "completions/min_length": 90.0, "completions/min_terminated_length": 90.0, "epoch": 0.816, "grad_norm": 0.0009517151629552245, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 856412304.0, "reward": 1.1400040626525878, "reward_std": 0.10449277609586716, "rewards/accuracy_reward": 0.48359375, "rewards/brier_reward": 0.796417486667633, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4969150543212891, "rewards/smce": -0.041142824292182925, "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": 0.0001953125, "completions/max_length": 907.2, "completions/max_terminated_length": 500.8, "completions/mean_length": 193.9853515625, "completions/mean_terminated_length": 193.72337951660157, "completions/min_length": 95.8, "completions/min_terminated_length": 95.8, "epoch": 0.832, "grad_norm": 0.0006846592877991498, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 873349946.0, "reward": 1.1227067232131958, "reward_std": 0.10306735038757324, "rewards/accuracy_reward": 0.44794921875, "rewards/brier_reward": 0.7976632714271545, "rewards/confidence_one_or_zero": 0.001171875, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4825107455253601, "rewards/smce": -0.04679301045835018, "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": 0.0001953125, "completions/max_length": 886.4, "completions/max_terminated_length": 466.2, "completions/mean_length": 199.3931640625, "completions/mean_terminated_length": 199.1320373535156, "completions/min_length": 92.2, "completions/min_terminated_length": 92.2, "epoch": 0.848, "grad_norm": 0.0008242399781011045, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 890447252.0, "reward": 1.1360819339752197, "reward_std": 0.08981756120920181, "rewards/accuracy_reward": 0.4703125, "rewards/brier_reward": 0.8020532727241516, "rewards/confidence_one_or_zero": 0.00244140625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.41491699814796446, "rewards/smce": -0.07371992468833924, "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": 720.0, "completions/max_terminated_length": 512.4, "completions/mean_length": 201.26787109375, "completions/mean_terminated_length": 201.13762817382812, "completions/min_length": 99.0, "completions/min_terminated_length": 99.0, "epoch": 0.864, "grad_norm": 0.0007919904310256243, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 907681771.0, "reward": 1.1619311332702638, "reward_std": 0.08479854315519333, "rewards/accuracy_reward": 0.52314453125, "rewards/brier_reward": 0.8008243322372437, "rewards/confidence_one_or_zero": 0.001171875, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4348466873168945, "rewards/smce": -0.097522396966815, "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": 0.00029296875, "completions/max_length": 1168.4, "completions/max_terminated_length": 544.4, "completions/mean_length": 203.99453125, "completions/mean_terminated_length": 203.60350646972657, "completions/min_length": 96.4, "completions/min_terminated_length": 96.4, "epoch": 0.88, "grad_norm": 0.0009532640688121319, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 924848563.0, "reward": 1.16089608669281, "reward_std": 0.10618025660514832, "rewards/accuracy_reward": 0.52587890625, "rewards/brier_reward": 0.7962094664573669, "rewards/confidence_one_or_zero": 0.00107421875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5438671946525574, "rewards/smce": -0.04220922961831093, "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": 708.0, "completions/max_terminated_length": 502.2, "completions/mean_length": 209.7365234375, "completions/mean_terminated_length": 209.60684814453126, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.896, "grad_norm": 0.0010002513881772757, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 941986857.0, "reward": 1.1198976755142211, "reward_std": 0.11406386941671372, "rewards/accuracy_reward": 0.4515625, "rewards/brier_reward": 0.7884399652481079, "rewards/confidence_one_or_zero": 0.00078125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5795322299003601, "rewards/smce": -0.12874267250299454, "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": 0.0, "completions/max_length": 551.0, "completions/max_terminated_length": 551.0, "completions/mean_length": 207.69072265625, "completions/mean_terminated_length": 207.69072265625, "completions/min_length": 105.2, "completions/min_terminated_length": 105.2, "epoch": 0.912, "grad_norm": 0.0007915336755104363, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 959217642.0, "reward": 1.1549952268600463, "reward_std": 0.10957214832305909, "rewards/accuracy_reward": 0.5193359375, "rewards/brier_reward": 0.7906574726104736, "rewards/confidence_one_or_zero": 0.00146484375, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5557373106479645, "rewards/smce": -0.0403962416574359, "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": 0.0, "completions/max_length": 558.0, "completions/max_terminated_length": 558.0, "completions/mean_length": 215.7451171875, "completions/mean_terminated_length": 215.7451171875, "completions/min_length": 113.4, "completions/min_terminated_length": 113.4, "epoch": 0.928, "grad_norm": 0.0009195299935527146, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 976234712.0, "reward": 1.1424710273742675, "reward_std": 0.10220244973897934, "rewards/accuracy_reward": 0.47255859375, "rewards/brier_reward": 0.8123879432678223, "rewards/confidence_one_or_zero": 0.00302734375, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4893798828125, "rewards/smce": -0.05421997383236885, "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": 0.0001953125, "completions/max_length": 1016.4, "completions/max_terminated_length": 618.6, "completions/mean_length": 226.06904296875, "completions/mean_terminated_length": 225.81295776367188, "completions/min_length": 113.4, "completions/min_terminated_length": 113.4, "epoch": 0.944, "grad_norm": 0.0008430502493865788, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 993718171.0, "reward": 1.1584136962890625, "reward_std": 0.08734249025583267, "rewards/accuracy_reward": 0.50771484375, "rewards/brier_reward": 0.8093124866485596, "rewards/confidence_one_or_zero": 0.00087890625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.45402538776397705, "rewards/smce": -0.05462546311318874, "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": 0.0001953125, "completions/max_length": 946.4, "completions/max_terminated_length": 563.4, "completions/mean_length": 231.7501953125, "completions/mean_terminated_length": 231.4956481933594, "completions/min_length": 109.2, "completions/min_terminated_length": 109.2, "epoch": 0.96, "grad_norm": 0.0007768862997181714, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 1011168445.0, "reward": 1.1429346323013305, "reward_std": 0.10092819035053253, "rewards/accuracy_reward": 0.48017578125, "rewards/brier_reward": 0.8058898329734803, "rewards/confidence_one_or_zero": 0.00302734375, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4817968726158142, "rewards/smce": -0.019481103867292404, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 460.5, "eval_completions/max_terminated_length": 460.5, "eval_completions/mean_length": 238.47315979003906, "eval_completions/mean_terminated_length": 238.47315979003906, "eval_completions/min_length": 115.0, "eval_completions/min_terminated_length": 115.0, "eval_loss": 0.0, "eval_num_tokens": 1011168445.0, "eval_reward": 1.0985642075538635, "eval_reward_std": 0.27343614399433136, "eval_rewards/accuracy_reward": 0.38671875, "eval_rewards/brier_reward": 0.8104199469089508, "eval_rewards/confidence_one_or_zero": 0.001953125, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4990234375, "eval_rewards/smce": -0.11257324367761612, "eval_runtime": 14.2596, "eval_samples_per_second": 35.064, "eval_steps_per_second": 0.14, "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": 0.00029296875, "completions/max_length": 1123.8, "completions/max_terminated_length": 628.0, "completions/mean_length": 237.015234375, "completions/mean_terminated_length": 236.63544921875, "completions/min_length": 118.8, "completions/min_terminated_length": 118.8, "epoch": 0.976, "grad_norm": 0.0008644487243145704, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 1028736857.0, "reward": 1.1499141216278077, "reward_std": 0.0988913893699646, "rewards/accuracy_reward": 0.47451171875, "rewards/brier_reward": 0.8256130218505859, "rewards/confidence_one_or_zero": 0.00322265625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5035009860992432, "rewards/smce": -0.044948733597993853, "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": 9.765625e-05, "completions/max_length": 892.0, "completions/max_terminated_length": 709.0, "completions/mean_length": 235.7791015625, "completions/mean_terminated_length": 235.65211791992186, "completions/min_length": 118.2, "completions/min_terminated_length": 118.2, "epoch": 0.992, "grad_norm": 0.000783053576014936, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 1046055459.0, "reward": 1.1583145380020141, "reward_std": 0.09551425129175187, "rewards/accuracy_reward": 0.52236328125, "rewards/brier_reward": 0.794464111328125, "rewards/confidence_one_or_zero": 0.00166015625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5242343783378601, "rewards/smce": -0.040295251086354254, "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": 0.00048828125, "completions/max_length": 1042.5, "completions/max_terminated_length": 762.5, "completions/mean_length": 237.88067626953125, "completions/mean_terminated_length": 237.24916076660156, "completions/min_length": 113.0, "completions/min_terminated_length": 113.0, "epoch": 0.9984, "num_tokens": 1053120646.0, "reward": 1.1611324548721313, "reward_std": 0.10324221849441528, "rewards/accuracy_reward": 0.515380859375, "rewards/brier_reward": 0.807620108127594, "rewards/confidence_one_or_zero": 0.0009765625, "rewards/format_reward": 0.999267578125, "rewards/mean_confidence_reward": 0.4963721036911011, "rewards/smce": -0.04539550840854645, "step": 312, "total_flos": 0.0, "train_loss": 0.00408483054268753, "train_runtime": 42881.0261, "train_samples_per_second": 0.466, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1053120646, "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 }