{ "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": 0.10751953125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1533.4, "completions/mean_length": 467.5353515625, "completions/mean_terminated_length": 338.8080322265625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.00392949627712369, "learning_rate": 3.1249999999999997e-07, "loss": 0.049, "num_tokens": 19631594.0, "reward": 0.2172173887491226, "reward_std": 0.39361864924430845, "rewards/accuracy_reward": 0.0654296875, "rewards/brier_reward": 0.11202050149440765, "rewards/confidence_one_or_zero": 0.4240234375, "rewards/format_reward": 0.26181640625, "rewards/mean_confidence_reward": 0.5485599517822266, "rewards/smce": -0.4833765268325806, "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": 0.1005859375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1530.6, "completions/mean_length": 442.7353515625, "completions/mean_terminated_length": 320.555126953125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.0044568004086613655, "learning_rate": 6.249999999999999e-07, "loss": 0.0491, "num_tokens": 39265556.0, "reward": 0.2336795836687088, "reward_std": 0.392722350358963, "rewards/accuracy_reward": 0.06435546875, "rewards/brier_reward": 0.1137367457151413, "rewards/confidence_one_or_zero": 0.436328125, "rewards/format_reward": 0.2943359375, "rewards/mean_confidence_reward": 0.5709773063659668, "rewards/smce": -0.5070997059345246, "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": 0.05361328125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1515.0, "completions/mean_length": 311.44423828125, "completions/mean_terminated_length": 242.5269287109375, "completions/min_length": 1.6, "completions/min_terminated_length": 1.6, "epoch": 0.048, "grad_norm": 0.0040215696208179, "learning_rate": 9.374999999999999e-07, "loss": 0.0516, "num_tokens": 57503481.0, "reward": 0.3996890068054199, "reward_std": 0.43802899718284605, "rewards/accuracy_reward": 0.1064453125, "rewards/brier_reward": 0.1968934416770935, "rewards/confidence_one_or_zero": 0.53427734375, "rewards/format_reward": 0.501953125, "rewards/mean_confidence_reward": 0.6975669860839844, "rewards/smce": -0.5916327595710754, "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": 0.0208984375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1477.2, "completions/mean_length": 209.77099609375, "completions/mean_terminated_length": 181.59221801757812, "completions/min_length": 7.4, "completions/min_terminated_length": 7.4, "epoch": 0.064, "grad_norm": 0.0028011025860905647, "learning_rate": 1e-06, "loss": 0.0307, "num_tokens": 74569936.0, "reward": 0.6740051746368408, "reward_std": 0.3955989480018616, "rewards/accuracy_reward": 0.1849609375, "rewards/brier_reward": 0.33165050745010377, "rewards/confidence_one_or_zero": 0.5771484375, "rewards/format_reward": 0.83759765625, "rewards/mean_confidence_reward": 0.8032736539840698, "rewards/smce": -0.6201528906822205, "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": 0.01220703125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1322.0, "completions/mean_length": 160.39345703125, "completions/mean_terminated_length": 143.40494079589843, "completions/min_length": 20.4, "completions/min_terminated_length": 20.4, "epoch": 0.08, "grad_norm": 0.0025906760711222887, "learning_rate": 1e-06, "loss": 0.0177, "num_tokens": 91145517.0, "reward": 0.7873133897781373, "reward_std": 0.32279191017150877, "rewards/accuracy_reward": 0.2150390625, "rewards/brier_reward": 0.4018263161182404, "rewards/confidence_one_or_zero": 0.5064453125, "rewards/format_reward": 0.96357421875, "rewards/mean_confidence_reward": 0.7923147320747376, "rewards/smce": -0.581539797782898, "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": 0.00615234375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1209.2, "completions/mean_length": 143.08671875, "completions/mean_terminated_length": 134.46507263183594, "completions/min_length": 29.0, "completions/min_terminated_length": 29.0, "epoch": 0.096, "grad_norm": 0.0022409316152334213, "learning_rate": 1e-06, "loss": 0.0088, "num_tokens": 107655333.0, "reward": 0.8789462089538574, "reward_std": 0.28507915139198303, "rewards/accuracy_reward": 0.24111328125, "rewards/brier_reward": 0.5377443969249726, "rewards/confidence_one_or_zero": 0.32646484375, "rewards/format_reward": 0.983203125, "rewards/mean_confidence_reward": 0.6468424916267395, "rewards/smce": -0.4170338332653046, "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": 0.003515625, "completions/max_length": 1536.0, "completions/max_terminated_length": 959.0, "completions/mean_length": 134.5802734375, "completions/mean_terminated_length": 129.63559265136718, "completions/min_length": 39.4, "completions/min_terminated_length": 39.4, "epoch": 0.112, "grad_norm": 0.0014994166558608413, "learning_rate": 1e-06, "loss": 0.0071, "num_tokens": 124142907.0, "reward": 0.9858169555664062, "reward_std": 0.21316042244434358, "rewards/accuracy_reward": 0.278515625, "rewards/brier_reward": 0.7030979990959167, "rewards/confidence_one_or_zero": 0.1814453125, "rewards/format_reward": 0.991796875, "rewards/mean_confidence_reward": 0.42606642842292786, "rewards/smce": -0.17777334451675414, "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": 0.0015625, "completions/max_length": 1536.0, "completions/max_terminated_length": 761.2, "completions/mean_length": 130.01904296875, "completions/mean_terminated_length": 127.81943054199219, "completions/min_length": 43.2, "completions/min_terminated_length": 43.2, "epoch": 0.128, "grad_norm": 0.0010288904886692762, "learning_rate": 1e-06, "loss": 0.0027, "num_tokens": 140390974.0, "reward": 1.0206197261810304, "reward_std": 0.15273716747760774, "rewards/accuracy_reward": 0.2755859375, "rewards/brier_reward": 0.7700915813446045, "rewards/confidence_one_or_zero": 0.14150390625, "rewards/format_reward": 0.99619140625, "rewards/mean_confidence_reward": 0.3022731125354767, "rewards/smce": -0.06302693337202073, "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": 1236.8, "completions/max_terminated_length": 817.4, "completions/mean_length": 125.5994140625, "completions/mean_terminated_length": 124.77249755859376, "completions/min_length": 50.8, "completions/min_terminated_length": 50.8, "epoch": 0.144, "grad_norm": 0.0009289440931752324, "learning_rate": 1e-06, "loss": 0.0016, "num_tokens": 156627544.0, "reward": 1.0439924240112304, "reward_std": 0.12395856976509094, "rewards/accuracy_reward": 0.327734375, "rewards/brier_reward": 0.7630051136016845, "rewards/confidence_one_or_zero": 0.107421875, "rewards/format_reward": 0.99814453125, "rewards/mean_confidence_reward": 0.24927894175052642, "rewards/smce": -0.08998571038246155, "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": 0.00107421875, "completions/max_length": 1536.0, "completions/max_terminated_length": 507.2, "completions/mean_length": 126.9736328125, "completions/mean_terminated_length": 125.45684814453125, "completions/min_length": 55.4, "completions/min_terminated_length": 55.4, "epoch": 0.16, "grad_norm": 0.0008057853556238115, "learning_rate": 1e-06, "loss": 0.0026, "num_tokens": 172948682.0, "reward": 1.0449287652969361, "reward_std": 0.10962955206632614, "rewards/accuracy_reward": 0.32119140625, "rewards/brier_reward": 0.77177414894104, "rewards/confidence_one_or_zero": 0.0669921875, "rewards/format_reward": 0.99765625, "rewards/mean_confidence_reward": 0.2513626962900162, "rewards/smce": -0.07648447304964065, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 255.5, "eval_completions/max_terminated_length": 255.5, "eval_completions/mean_length": 123.6274299621582, "eval_completions/mean_terminated_length": 123.6274299621582, "eval_completions/min_length": 58.0, "eval_completions/min_terminated_length": 58.0, "eval_loss": 0.0, "eval_num_tokens": 172948682.0, "eval_reward": 1.021654725074768, "eval_reward_std": 0.1542477086186409, "eval_rewards/accuracy_reward": 0.23828125, "eval_rewards/brier_reward": 0.8073956966400146, "eval_rewards/confidence_one_or_zero": 0.07421875, "eval_rewards/format_reward": 0.998046875, "eval_rewards/mean_confidence_reward": 0.2683268338441849, "eval_rewards/smce": -0.04149917885661125, "eval_runtime": 10.0732, "eval_samples_per_second": 49.637, "eval_steps_per_second": 0.199, "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": 0.00029296875, "completions/max_length": 1093.0, "completions/max_terminated_length": 615.6, "completions/mean_length": 125.6130859375, "completions/mean_terminated_length": 125.19995574951172, "completions/min_length": 42.2, "completions/min_terminated_length": 42.2, "epoch": 0.176, "grad_norm": 0.000889862421900034, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 189472080.0, "reward": 1.045188045501709, "reward_std": 0.10808687955141068, "rewards/accuracy_reward": 0.313671875, "rewards/brier_reward": 0.7784487485885621, "rewards/confidence_one_or_zero": 0.037109375, "rewards/format_reward": 0.99873046875, "rewards/mean_confidence_reward": 0.27597495913505554, "rewards/smce": -0.04757043533027172, "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": 1110.2, "completions/max_terminated_length": 487.0, "completions/mean_length": 125.22353515625, "completions/mean_terminated_length": 124.67266235351562, "completions/min_length": 56.8, "completions/min_terminated_length": 56.8, "epoch": 0.192, "grad_norm": 0.0009008679771795869, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 205569185.0, "reward": 1.061819863319397, "reward_std": 0.11279134750366211, "rewards/accuracy_reward": 0.35380859375, "rewards/brier_reward": 0.7713178038597107, "rewards/confidence_one_or_zero": 0.016015625, "rewards/format_reward": 0.99912109375, "rewards/mean_confidence_reward": 0.3114722430706024, "rewards/smce": -0.06084534525871277, "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": 0.0001953125, "completions/max_length": 739.6, "completions/max_terminated_length": 579.0, "completions/mean_length": 128.28876953125, "completions/mean_terminated_length": 128.01407470703126, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.208, "grad_norm": 0.0007912637665867805, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 221915086.0, "reward": 1.0683193922042846, "reward_std": 0.11826074719429017, "rewards/accuracy_reward": 0.37177734375, "rewards/brier_reward": 0.7660857439041138, "rewards/confidence_one_or_zero": 0.0056640625, "rewards/format_reward": 0.99912109375, "rewards/mean_confidence_reward": 0.3447410583496094, "rewards/smce": -0.03460576832294464, "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": 341.8, "completions/max_terminated_length": 341.8, "completions/mean_length": 125.3783203125, "completions/mean_terminated_length": 125.3783203125, "completions/min_length": 58.8, "completions/min_terminated_length": 58.8, "epoch": 0.224, "grad_norm": 0.0009502115426585078, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 238352144.0, "reward": 1.068940782546997, "reward_std": 0.11054961085319519, "rewards/accuracy_reward": 0.3662109375, "rewards/brier_reward": 0.7722442865371704, "rewards/confidence_one_or_zero": 0.0056640625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.35045996904373167, "rewards/smce": -0.03791281208395958, "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": 0.0001953125, "completions/max_length": 806.0, "completions/max_terminated_length": 360.4, "completions/mean_length": 125.35654296875, "completions/mean_terminated_length": 125.08109130859376, "completions/min_length": 55.4, "completions/min_terminated_length": 55.4, "epoch": 0.24, "grad_norm": 0.0009807253954932094, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 254887475.0, "reward": 1.0809075117111206, "reward_std": 0.12458073198795319, "rewards/accuracy_reward": 0.405078125, "rewards/brier_reward": 0.7581897139549255, "rewards/confidence_one_or_zero": 0.00234375, "rewards/format_reward": 0.99921875, "rewards/mean_confidence_reward": 0.37911379933357237, "rewards/smce": -0.06723080649971962, "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": 595.6, "completions/max_terminated_length": 397.0, "completions/mean_length": 123.59150390625, "completions/mean_terminated_length": 123.45372619628907, "completions/min_length": 57.8, "completions/min_terminated_length": 57.8, "epoch": 0.256, "grad_norm": 0.0009870625799521804, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 271207868.0, "reward": 1.07213237285614, "reward_std": 0.1215924084186554, "rewards/accuracy_reward": 0.37939453125, "rewards/brier_reward": 0.7657799482345581, "rewards/confidence_one_or_zero": 0.0015625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4004322052001953, "rewards/smce": -0.06174885742366314, "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": 0.000390625, "completions/max_length": 1042.4, "completions/max_terminated_length": 344.4, "completions/mean_length": 126.02294921875, "completions/mean_terminated_length": 125.4724334716797, "completions/min_length": 60.4, "completions/min_terminated_length": 60.4, "epoch": 0.272, "grad_norm": 0.0008904941496439278, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 287464039.0, "reward": 1.0802291870117187, "reward_std": 0.11962677985429764, "rewards/accuracy_reward": 0.3966796875, "rewards/brier_reward": 0.7651029467582703, "rewards/confidence_one_or_zero": 0.001953125, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.3965742290019989, "rewards/smce": -0.0348443428054452, "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": 0.00029296875, "completions/max_length": 1114.2, "completions/max_terminated_length": 378.4, "completions/mean_length": 123.46328125, "completions/mean_terminated_length": 123.04900360107422, "completions/min_length": 57.8, "completions/min_terminated_length": 57.8, "epoch": 0.288, "grad_norm": 0.0010928181000053883, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 303686479.0, "reward": 1.079901361465454, "reward_std": 0.11479000747203827, "rewards/accuracy_reward": 0.39150390625, "rewards/brier_reward": 0.7692065358161926, "rewards/confidence_one_or_zero": 0.0017578125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.39766603112220766, "rewards/smce": -0.05178325399756432, "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": 347.8, "completions/max_terminated_length": 347.8, "completions/mean_length": 122.50859375, "completions/mean_terminated_length": 122.50859375, "completions/min_length": 58.0, "completions/min_terminated_length": 58.0, "epoch": 0.304, "grad_norm": 0.0009639639756642282, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 319870919.0, "reward": 1.087165641784668, "reward_std": 0.11266087889671325, "rewards/accuracy_reward": 0.40869140625, "rewards/brier_reward": 0.7662365555763244, "rewards/confidence_one_or_zero": 0.0009765625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3889293670654297, "rewards/smce": -0.04021448343992233, "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": 0.0001953125, "completions/max_length": 836.0, "completions/max_terminated_length": 372.6, "completions/mean_length": 123.7615234375, "completions/mean_terminated_length": 123.48605194091797, "completions/min_length": 57.0, "completions/min_terminated_length": 57.0, "epoch": 0.32, "grad_norm": 0.0008017583750188351, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 336226941.0, "reward": 1.0964327812194825, "reward_std": 0.10867121815681458, "rewards/accuracy_reward": 0.42861328125, "rewards/brier_reward": 0.7652536988258362, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3929861426353455, "rewards/smce": -0.07092485018074512, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 230.5, "eval_completions/max_terminated_length": 230.5, "eval_completions/mean_length": 123.81131362915039, "eval_completions/mean_terminated_length": 123.81131362915039, "eval_completions/min_length": 66.0, "eval_completions/min_terminated_length": 66.0, "eval_loss": 0.0, "eval_num_tokens": 336226941.0, "eval_reward": 1.0621264576911926, "eval_reward_std": 0.2113262265920639, "eval_rewards/accuracy_reward": 0.337890625, "eval_rewards/brier_reward": 0.7869812250137329, "eval_rewards/confidence_one_or_zero": 0.001953125, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.39699217677116394, "eval_rewards/smce": -0.06198457069694996, "eval_runtime": 9.5484, "eval_samples_per_second": 52.365, "eval_steps_per_second": 0.209, "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": 9.765625e-05, "completions/max_length": 559.2, "completions/max_terminated_length": 332.4, "completions/mean_length": 122.5482421875, "completions/mean_terminated_length": 122.41039276123047, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.336, "grad_norm": 0.000917735684197396, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 352204267.0, "reward": 1.093878197669983, "reward_std": 0.1106439396739006, "rewards/accuracy_reward": 0.42294921875, "rewards/brier_reward": 0.7657433271408081, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.40318635702133176, "rewards/smce": -0.054635743424296376, "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": 554.6, "completions/max_terminated_length": 328.8, "completions/mean_length": 126.69482421875, "completions/mean_terminated_length": 126.5573974609375, "completions/min_length": 60.4, "completions/min_terminated_length": 60.4, "epoch": 0.352, "grad_norm": 0.0008927257149480283, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 368762038.0, "reward": 1.0763854026794433, "reward_std": 0.10895265638828278, "rewards/accuracy_reward": 0.38125, "rewards/brier_reward": 0.7718688130378724, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4014307737350464, "rewards/smce": -0.02511356510221958, "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": 517.6, "completions/max_terminated_length": 517.6, "completions/mean_length": 132.65068359375, "completions/mean_terminated_length": 132.65068359375, "completions/min_length": 63.6, "completions/min_terminated_length": 63.6, "epoch": 0.368, "grad_norm": 0.000985518447123468, "learning_rate": 1e-06, "loss": -0.0006, "num_tokens": 385185853.0, "reward": 1.0864951610565186, "reward_std": 0.10731481909751892, "rewards/accuracy_reward": 0.3982421875, "rewards/brier_reward": 0.775156557559967, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4145812392234802, "rewards/smce": -0.04092612899839878, "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": 0.0001953125, "completions/max_length": 592.8, "completions/max_terminated_length": 356.0, "completions/mean_length": 138.587890625, "completions/mean_terminated_length": 138.31441650390624, "completions/min_length": 58.6, "completions/min_terminated_length": 58.6, "epoch": 0.384, "grad_norm": 0.0009577263845130801, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 401461505.0, "reward": 1.097441053390503, "reward_std": 0.11049059182405471, "rewards/accuracy_reward": 0.41982421875, "rewards/brier_reward": 0.7755703687667846, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4149980843067169, "rewards/smce": -0.03180096782743931, "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": 0.0, "completions/max_length": 444.4, "completions/max_terminated_length": 444.4, "completions/mean_length": 145.02041015625, "completions/mean_terminated_length": 145.02041015625, "completions/min_length": 65.2, "completions/min_terminated_length": 65.2, "epoch": 0.4, "grad_norm": 0.0009062528843060136, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 417982962.0, "reward": 1.0899156332015991, "reward_std": 0.11331143081188202, "rewards/accuracy_reward": 0.39873046875, "rewards/brier_reward": 0.7814009428024292, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.41860421299934386, "rewards/smce": -0.030102025158703326, "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": 867.0, "completions/max_terminated_length": 467.8, "completions/mean_length": 153.2337890625, "completions/mean_terminated_length": 152.82854614257812, "completions/min_length": 65.4, "completions/min_terminated_length": 65.4, "epoch": 0.416, "grad_norm": 0.0009052875684574246, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 434433260.0, "reward": 1.0993643522262573, "reward_std": 0.11355077922344207, "rewards/accuracy_reward": 0.4205078125, "rewards/brier_reward": 0.7787749409675598, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4114785134792328, "rewards/smce": -0.026191336289048194, "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": 0.00029296875, "completions/max_length": 1097.6, "completions/max_terminated_length": 478.6, "completions/mean_length": 158.43994140625, "completions/mean_terminated_length": 158.03628845214843, "completions/min_length": 69.6, "completions/min_terminated_length": 69.6, "epoch": 0.432, "grad_norm": 0.0008921238477341831, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 451070021.0, "reward": 1.1083012580871583, "reward_std": 0.10892681926488876, "rewards/accuracy_reward": 0.44169921875, "rewards/brier_reward": 0.7755492091178894, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4254137694835663, "rewards/smce": -0.03537834882736206, "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": 0.00029296875, "completions/max_length": 857.0, "completions/max_terminated_length": 403.8, "completions/mean_length": 162.83359375, "completions/mean_terminated_length": 162.4312957763672, "completions/min_length": 64.2, "completions/min_terminated_length": 64.2, "epoch": 0.448, "grad_norm": 0.0008144128369167447, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 467690237.0, "reward": 1.0995479583740235, "reward_std": 0.11465191096067429, "rewards/accuracy_reward": 0.42607421875, "rewards/brier_reward": 0.7738674759864808, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.44896326065063474, "rewards/smce": -0.0358337426558137, "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": 0.00048828125, "completions/max_length": 1316.2, "completions/max_terminated_length": 476.2, "completions/mean_length": 164.81923828125, "completions/mean_terminated_length": 164.14944458007812, "completions/min_length": 75.0, "completions/min_terminated_length": 75.0, "epoch": 0.464, "grad_norm": 0.0008099125698208809, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 484548802.0, "reward": 1.0764792203903197, "reward_std": 0.11227991878986358, "rewards/accuracy_reward": 0.37109375, "rewards/brier_reward": 0.7830622434616089, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.4407523512840271, "rewards/smce": -0.0710053825750947, "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": 9.765625e-05, "completions/max_length": 665.4, "completions/max_terminated_length": 441.8, "completions/mean_length": 162.240625, "completions/mean_terminated_length": 162.10667419433594, "completions/min_length": 76.8, "completions/min_terminated_length": 76.8, "epoch": 0.48, "grad_norm": 0.000975205737631768, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 501258178.0, "reward": 1.1072642087936402, "reward_std": 0.10939528942108154, "rewards/accuracy_reward": 0.44150390625, "rewards/brier_reward": 0.7736056089401245, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4199238717556, "rewards/smce": -0.04843057133257389, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 317.5, "eval_completions/max_terminated_length": 317.5, "eval_completions/mean_length": 163.09646606445312, "eval_completions/mean_terminated_length": 163.09646606445312, "eval_completions/min_length": 79.0, "eval_completions/min_terminated_length": 79.0, "eval_loss": 0.0, "eval_num_tokens": 501258178.0, "eval_reward": 1.073753833770752, "eval_reward_std": 0.21566922962665558, "eval_rewards/accuracy_reward": 0.35546875, "eval_rewards/brier_reward": 0.7925025522708893, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.40083983540534973, "eval_rewards/smce": -0.04645352065563202, "eval_runtime": 11.8477, "eval_samples_per_second": 42.202, "eval_steps_per_second": 0.169, "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": 678.4, "completions/max_terminated_length": 462.0, "completions/mean_length": 160.70107421875, "completions/mean_terminated_length": 160.56687927246094, "completions/min_length": 72.2, "completions/min_terminated_length": 72.2, "epoch": 0.496, "grad_norm": 0.0009712612954899669, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 518211597.0, "reward": 1.1064111471176148, "reward_std": 0.10755210518836975, "rewards/accuracy_reward": 0.44091796875, "rewards/brier_reward": 0.7725687384605407, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.408827143907547, "rewards/smce": -0.037215230613946916, "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": 0.0001953125, "completions/max_length": 699.4, "completions/max_terminated_length": 468.8, "completions/mean_length": 160.16220703125, "completions/mean_terminated_length": 159.8941864013672, "completions/min_length": 73.0, "completions/min_terminated_length": 73.0, "epoch": 0.512, "grad_norm": 0.0009234827593900263, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 534997322.0, "reward": 1.1220848321914674, "reward_std": 0.10917693674564362, "rewards/accuracy_reward": 0.4650390625, "rewards/brier_reward": 0.7797863245010376, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.42907978892326354, "rewards/smce": -0.04612938240170479, "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": 0.0, "completions/max_length": 487.0, "completions/max_terminated_length": 487.0, "completions/mean_length": 164.23193359375, "completions/mean_terminated_length": 164.23193359375, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.528, "grad_norm": 0.0009526491048745811, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 551708593.0, "reward": 1.113792848587036, "reward_std": 0.11661270409822463, "rewards/accuracy_reward": 0.4375, "rewards/brier_reward": 0.7903852462768555, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4550130248069763, "rewards/smce": -0.030036850459873678, "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": 652.8, "completions/max_terminated_length": 450.2, "completions/mean_length": 162.72724609375, "completions/mean_terminated_length": 162.5930206298828, "completions/min_length": 71.0, "completions/min_terminated_length": 71.0, "epoch": 0.544, "grad_norm": 0.0009181004716083407, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 568538504.0, "reward": 1.1319088459014892, "reward_std": 0.12556345760822296, "rewards/accuracy_reward": 0.490234375, "rewards/brier_reward": 0.7738756775856018, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.48188769817352295, "rewards/smce": -0.019567849487066268, "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": 0.0, "completions/max_length": 440.4, "completions/max_terminated_length": 440.4, "completions/mean_length": 158.8845703125, "completions/mean_terminated_length": 158.8845703125, "completions/min_length": 70.4, "completions/min_terminated_length": 70.4, "epoch": 0.56, "grad_norm": 0.0008448117878288031, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 584986890.0, "reward": 1.1044183254241944, "reward_std": 0.11335271000862121, "rewards/accuracy_reward": 0.419921875, "rewards/brier_reward": 0.7894919633865356, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.476855480670929, "rewards/smce": -0.057809342816472056, "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": 412.2, "completions/max_terminated_length": 412.2, "completions/mean_length": 158.0970703125, "completions/mean_terminated_length": 158.0970703125, "completions/min_length": 73.6, "completions/min_terminated_length": 73.6, "epoch": 0.576, "grad_norm": 0.0007008586544543505, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 601792428.0, "reward": 1.1063513040542603, "reward_std": 0.10015368312597275, "rewards/accuracy_reward": 0.42802734375, "rewards/brier_reward": 0.7851521253585816, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4443991184234619, "rewards/smce": -0.047779816389083865, "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": 0.0001953125, "completions/max_length": 655.8, "completions/max_terminated_length": 417.0, "completions/mean_length": 154.9142578125, "completions/mean_terminated_length": 154.64378356933594, "completions/min_length": 74.6, "completions/min_terminated_length": 74.6, "epoch": 0.592, "grad_norm": 0.0007284689345397055, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 618546462.0, "reward": 1.1119030475616456, "reward_std": 0.09350670576095581, "rewards/accuracy_reward": 0.44013671875, "rewards/brier_reward": 0.7842747569084167, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4142773449420929, "rewards/smce": -0.03132548592984676, "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": 0.0, "completions/max_length": 424.8, "completions/max_terminated_length": 424.8, "completions/mean_length": 154.5126953125, "completions/mean_terminated_length": 154.5126953125, "completions/min_length": 74.4, "completions/min_terminated_length": 74.4, "epoch": 0.608, "grad_norm": 0.0008701604674570262, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 635128160.0, "reward": 1.1218822956085206, "reward_std": 0.10320815443992615, "rewards/accuracy_reward": 0.44443359375, "rewards/brier_reward": 0.7996853113174438, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.43260841369628905, "rewards/smce": -0.035519332997500895, "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": 618.2, "completions/max_terminated_length": 392.8, "completions/mean_length": 157.1732421875, "completions/mean_terminated_length": 157.0383514404297, "completions/min_length": 69.4, "completions/min_terminated_length": 69.4, "epoch": 0.624, "grad_norm": 0.0011468639131635427, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 652081518.0, "reward": 1.1084702968597413, "reward_std": 0.1136349618434906, "rewards/accuracy_reward": 0.4287109375, "rewards/brier_reward": 0.7887693285942078, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4634462893009186, "rewards/smce": -0.044282649829983714, "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": 0.00029296875, "completions/max_length": 1093.6, "completions/max_terminated_length": 420.0, "completions/mean_length": 157.331640625, "completions/mean_terminated_length": 156.92776794433593, "completions/min_length": 74.6, "completions/min_terminated_length": 74.6, "epoch": 0.64, "grad_norm": 0.0008401005179621279, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 669035282.0, "reward": 1.143984031677246, "reward_std": 0.11052749902009965, "rewards/accuracy_reward": 0.49716796875, "rewards/brier_reward": 0.7917260885238647, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4753476619720459, "rewards/smce": -0.05361993685364723, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 320.0, "eval_completions/max_terminated_length": 320.0, "eval_completions/mean_length": 161.03005981445312, "eval_completions/mean_terminated_length": 161.03005981445312, "eval_completions/min_length": 83.5, "eval_completions/min_terminated_length": 83.5, "eval_loss": 0.0, "eval_num_tokens": 669035282.0, "eval_reward": 1.0816600322723389, "eval_reward_std": 0.2524609863758087, "eval_rewards/accuracy_reward": 0.3671875, "eval_rewards/brier_reward": 0.7971554696559906, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.46910156309604645, "eval_rewards/smce": -0.10237480700016022, "eval_runtime": 12.0043, "eval_samples_per_second": 41.652, "eval_steps_per_second": 0.167, "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": 9.765625e-05, "completions/max_length": 623.4, "completions/max_terminated_length": 380.4, "completions/mean_length": 157.94111328125, "completions/mean_terminated_length": 157.80642395019532, "completions/min_length": 75.6, "completions/min_terminated_length": 75.6, "epoch": 0.656, "grad_norm": 0.001043739845044911, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 685509143.0, "reward": 1.1010662317276, "reward_std": 0.11220352202653885, "rewards/accuracy_reward": 0.417578125, "rewards/brier_reward": 0.7853797674179077, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.47954005002975464, "rewards/smce": -0.0630951227620244, "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": 612.2, "completions/max_terminated_length": 395.2, "completions/mean_length": 158.19736328125, "completions/mean_terminated_length": 158.0628631591797, "completions/min_length": 73.2, "completions/min_terminated_length": 73.2, "epoch": 0.672, "grad_norm": 0.0007125500706024468, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 702042524.0, "reward": 1.1255898952484131, "reward_std": 0.09300058484077453, "rewards/accuracy_reward": 0.45703125, "rewards/brier_reward": 0.7946899771690369, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4336543083190918, "rewards/smce": -0.04446049779653549, "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": 434.0, "completions/max_terminated_length": 434.0, "completions/mean_length": 158.30068359375, "completions/mean_terminated_length": 158.30068359375, "completions/min_length": 74.0, "completions/min_terminated_length": 74.0, "epoch": 0.688, "grad_norm": 0.0009905676124617457, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 718617443.0, "reward": 1.1272786140441895, "reward_std": 0.09988767951726914, "rewards/accuracy_reward": 0.45419921875, "rewards/brier_reward": 0.8008356928825379, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.42675782442092897, "rewards/smce": -0.04784578718245029, "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": 0.0, "completions/max_length": 399.4, "completions/max_terminated_length": 399.4, "completions/mean_length": 160.6234375, "completions/mean_terminated_length": 160.6234375, "completions/min_length": 75.4, "completions/min_terminated_length": 75.4, "epoch": 0.704, "grad_norm": 0.0008583047892898321, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 735128371.0, "reward": 1.1382115125656127, "reward_std": 0.1009350448846817, "rewards/accuracy_reward": 0.48193359375, "rewards/brier_reward": 0.7947717547416687, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.46465528607368467, "rewards/smce": -0.028317653015255927, "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": 0.0001953125, "completions/max_length": 879.0, "completions/max_terminated_length": 483.4, "completions/mean_length": 164.634375, "completions/mean_terminated_length": 164.36683349609376, "completions/min_length": 64.8, "completions/min_terminated_length": 64.8, "epoch": 0.72, "grad_norm": 0.0009704248514026403, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 751824083.0, "reward": 1.1468546390533447, "reward_std": 0.10621728748083115, "rewards/accuracy_reward": 0.4880859375, "rewards/brier_reward": 0.8062491059303284, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5047011613845825, "rewards/smce": -0.033376603946089746, "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": 858.4, "completions/max_terminated_length": 529.8, "completions/mean_length": 165.065625, "completions/mean_terminated_length": 164.79816284179688, "completions/min_length": 77.0, "completions/min_terminated_length": 77.0, "epoch": 0.736, "grad_norm": 0.0009971307590603828, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 768453939.0, "reward": 1.1445934057235718, "reward_std": 0.10860948264598846, "rewards/accuracy_reward": 0.49833984375, "rewards/brier_reward": 0.7916695952415467, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5039228439331055, "rewards/smce": -0.053068242594599725, "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": 0.0, "completions/max_length": 449.0, "completions/max_terminated_length": 449.0, "completions/mean_length": 169.20380859375, "completions/mean_terminated_length": 169.20380859375, "completions/min_length": 80.2, "completions/min_terminated_length": 80.2, "epoch": 0.752, "grad_norm": 0.0009284823318012059, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 785413786.0, "reward": 1.1324732303619385, "reward_std": 0.10707245618104935, "rewards/accuracy_reward": 0.47109375, "rewards/brier_reward": 0.7942162990570069, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4771093845367432, "rewards/smce": -0.036446777358651164, "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": 874.0, "completions/max_terminated_length": 661.4, "completions/mean_length": 175.115625, "completions/mean_terminated_length": 174.98184814453126, "completions/min_length": 74.6, "completions/min_terminated_length": 74.6, "epoch": 0.768, "grad_norm": 0.0008781220531091094, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 802139674.0, "reward": 1.1389093160629273, "reward_std": 0.09807263612747193, "rewards/accuracy_reward": 0.473046875, "rewards/brier_reward": 0.8051334381103515, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.46835254430770873, "rewards/smce": -0.016733316145837306, "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": 0.0, "completions/max_length": 418.6, "completions/max_terminated_length": 418.6, "completions/mean_length": 171.97890625, "completions/mean_terminated_length": 171.97890625, "completions/min_length": 84.4, "completions/min_terminated_length": 84.4, "epoch": 0.784, "grad_norm": 0.0009764548158273101, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 819075106.0, "reward": 1.1391575574874877, "reward_std": 0.10245644003152847, "rewards/accuracy_reward": 0.48662109375, "rewards/brier_reward": 0.7924958348274231, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.47624804377555846, "rewards/smce": -0.08027197942137718, "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": 485.6, "completions/max_terminated_length": 485.6, "completions/mean_length": 169.80869140625, "completions/mean_terminated_length": 169.80869140625, "completions/min_length": 84.6, "completions/min_terminated_length": 84.6, "epoch": 0.8, "grad_norm": 0.0011034056078642607, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 835824507.0, "reward": 1.1558006286621094, "reward_std": 0.11075237542390823, "rewards/accuracy_reward": 0.50732421875, "rewards/brier_reward": 0.8045778751373291, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5076025426387787, "rewards/smce": -0.030179687216877936, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 340.5, "eval_completions/max_terminated_length": 340.5, "eval_completions/mean_length": 174.92011260986328, "eval_completions/mean_terminated_length": 174.92011260986328, "eval_completions/min_length": 89.5, "eval_completions/min_terminated_length": 89.5, "eval_loss": 0.0, "eval_num_tokens": 835824507.0, "eval_reward": 1.0990996360778809, "eval_reward_std": 0.2727942317724228, "eval_rewards/accuracy_reward": 0.40234375, "eval_rewards/brier_reward": 0.7968310713768005, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.49873048067092896, "eval_rewards/smce": -0.09765136614441872, "eval_runtime": 11.7282, "eval_samples_per_second": 42.632, "eval_steps_per_second": 0.171, "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": 0.0001953125, "completions/max_length": 686.8, "completions/max_terminated_length": 486.0, "completions/mean_length": 171.3958984375, "completions/mean_terminated_length": 171.12869567871093, "completions/min_length": 88.0, "completions/min_terminated_length": 88.0, "epoch": 0.816, "grad_norm": 0.0011392795713618398, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 852678769.0, "reward": 1.1497153043746948, "reward_std": 0.114381605386734, "rewards/accuracy_reward": 0.494921875, "rewards/brier_reward": 0.8050509095191956, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5240488409996032, "rewards/smce": -0.03478394001722336, "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": 0.0, "completions/max_length": 498.0, "completions/max_terminated_length": 498.0, "completions/mean_length": 176.65927734375, "completions/mean_terminated_length": 176.65927734375, "completions/min_length": 86.6, "completions/min_terminated_length": 86.6, "epoch": 0.832, "grad_norm": 0.001069263438694179, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 869496112.0, "reward": 1.1582709312438966, "reward_std": 0.10491050034761429, "rewards/accuracy_reward": 0.50341796875, "rewards/brier_reward": 0.8133532643318176, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5025537252426148, "rewards/smce": -0.023029053397476674, "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": 0.0, "completions/max_length": 502.4, "completions/max_terminated_length": 502.4, "completions/mean_length": 176.24482421875, "completions/mean_terminated_length": 176.24482421875, "completions/min_length": 85.4, "completions/min_terminated_length": 85.4, "epoch": 0.848, "grad_norm": 0.0009353526984341443, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 886315227.0, "reward": 1.139298939704895, "reward_std": 0.10831455588340759, "rewards/accuracy_reward": 0.47412109375, "rewards/brier_reward": 0.804800283908844, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4926269590854645, "rewards/smce": -0.03244311548769474, "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": 0.0, "completions/max_length": 494.6, "completions/max_terminated_length": 494.6, "completions/mean_length": 179.4521484375, "completions/mean_terminated_length": 179.4521484375, "completions/min_length": 89.4, "completions/min_terminated_length": 89.4, "epoch": 0.864, "grad_norm": 0.0009035305120050907, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 903139633.0, "reward": 1.1543693780899047, "reward_std": 0.10487027764320374, "rewards/accuracy_reward": 0.50732421875, "rewards/brier_reward": 0.8016931176185608, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49616700410842896, "rewards/smce": -0.02795556578785181, "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": 0.000390625, "completions/max_length": 1336.4, "completions/max_terminated_length": 506.0, "completions/mean_length": 182.682421875, "completions/mean_terminated_length": 182.15309143066406, "completions/min_length": 91.8, "completions/min_terminated_length": 91.8, "epoch": 0.88, "grad_norm": 0.0007824220811016858, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 920157373.0, "reward": 1.1384138345718384, "reward_std": 0.09232062548398971, "rewards/accuracy_reward": 0.46337890625, "rewards/brier_reward": 0.8143354654312134, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4651464939117432, "rewards/smce": -0.04969409257173538, "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": 0.0, "completions/max_length": 473.0, "completions/max_terminated_length": 473.0, "completions/mean_length": 183.71357421875, "completions/mean_terminated_length": 183.71357421875, "completions/min_length": 92.2, "completions/min_terminated_length": 92.2, "epoch": 0.896, "grad_norm": 0.0010242619318887591, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 937149448.0, "reward": 1.143228530883789, "reward_std": 0.09831419289112091, "rewards/accuracy_reward": 0.47509765625, "rewards/brier_reward": 0.8117444038391113, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4801484525203705, "rewards/smce": -0.03859155103564262, "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": 474.8, "completions/max_terminated_length": 474.8, "completions/mean_length": 181.05634765625, "completions/mean_terminated_length": 181.05634765625, "completions/min_length": 88.0, "completions/min_terminated_length": 88.0, "epoch": 0.912, "grad_norm": 0.0009002075530588627, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 954054761.0, "reward": 1.1372110366821289, "reward_std": 0.10027643889188767, "rewards/accuracy_reward": 0.46640625, "rewards/brier_reward": 0.8085625290870666, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49353516697883604, "rewards/smce": -0.05476636029779911, "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": 532.6, "completions/max_terminated_length": 532.6, "completions/mean_length": 174.0798828125, "completions/mean_terminated_length": 174.0798828125, "completions/min_length": 90.2, "completions/min_terminated_length": 90.2, "epoch": 0.928, "grad_norm": 0.0008942595450207591, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 970864155.0, "reward": 1.1491928815841674, "reward_std": 0.09633661955595016, "rewards/accuracy_reward": 0.50048828125, "rewards/brier_reward": 0.7983452439308166, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4858271420001984, "rewards/smce": -0.044872155785560607, "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": 9.765625e-05, "completions/max_length": 682.2, "completions/max_terminated_length": 449.4, "completions/mean_length": 171.64189453125, "completions/mean_terminated_length": 171.50814208984374, "completions/min_length": 85.8, "completions/min_terminated_length": 85.8, "epoch": 0.944, "grad_norm": 0.0009355693473480642, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 987597192.0, "reward": 1.1451186180114745, "reward_std": 0.0981968179345131, "rewards/accuracy_reward": 0.4908203125, "rewards/brier_reward": 0.8003330111503602, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47065430879592896, "rewards/smce": -0.08194140195846558, "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": 506.2, "completions/max_terminated_length": 506.2, "completions/mean_length": 172.3828125, "completions/mean_terminated_length": 172.3828125, "completions/min_length": 86.6, "completions/min_terminated_length": 86.6, "epoch": 0.96, "grad_norm": 0.0008330916170962155, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 1004302712.0, "reward": 1.1374672889709472, "reward_std": 0.08538994491100312, "rewards/accuracy_reward": 0.46298828125, "rewards/brier_reward": 0.8122195124626159, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.46052245497703553, "rewards/smce": -0.027407228574156763, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 349.5, "eval_completions/max_terminated_length": 349.5, "eval_completions/mean_length": 173.41207122802734, "eval_completions/mean_terminated_length": 173.41207122802734, "eval_completions/min_length": 98.0, "eval_completions/min_terminated_length": 98.0, "eval_loss": 0.0, "eval_num_tokens": 1004302712.0, "eval_reward": 1.1020442247390747, "eval_reward_std": 0.2523525282740593, "eval_rewards/accuracy_reward": 0.388671875, "eval_rewards/brier_reward": 0.8159765899181366, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4429687410593033, "eval_rewards/smce": -0.05609375610947609, "eval_runtime": 12.0381, "eval_samples_per_second": 41.535, "eval_steps_per_second": 0.166, "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": 9.765625e-05, "completions/max_length": 707.2, "completions/max_terminated_length": 491.8, "completions/mean_length": 172.77578125, "completions/mean_terminated_length": 172.642529296875, "completions/min_length": 90.6, "completions/min_terminated_length": 90.6, "epoch": 0.976, "grad_norm": 0.0008364045643247664, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 1020933056.0, "reward": 1.1579480171203613, "reward_std": 0.08691856116056443, "rewards/accuracy_reward": 0.50361328125, "rewards/brier_reward": 0.8128007888793946, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4667675793170929, "rewards/smce": -0.04212206564843655, "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": 0.0, "completions/max_length": 612.8, "completions/max_terminated_length": 612.8, "completions/mean_length": 171.72080078125, "completions/mean_terminated_length": 171.72080078125, "completions/min_length": 86.8, "completions/min_terminated_length": 86.8, "epoch": 0.992, "grad_norm": 0.0009284996194764972, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 1037819957.0, "reward": 1.1416526317596436, "reward_std": 0.10318480283021927, "rewards/accuracy_reward": 0.4888671875, "rewards/brier_reward": 0.7947907686233521, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4984033167362213, "rewards/smce": -0.03536987099796533, "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": 0.000244140625, "completions/max_length": 956.0, "completions/max_terminated_length": 389.0, "completions/mean_length": 169.3539276123047, "completions/mean_terminated_length": 169.02008056640625, "completions/min_length": 84.5, "completions/min_terminated_length": 84.5, "epoch": 0.9984, "num_tokens": 1044519143.0, "reward": 1.129215657711029, "reward_std": 0.10631396621465683, "rewards/accuracy_reward": 0.47412109375, "rewards/brier_reward": 0.7854089140892029, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999755859375, "rewards/mean_confidence_reward": 0.5580566227436066, "rewards/smce": -0.08556640334427357, "step": 312, "total_flos": 0.0, "train_loss": 0.003961190384311171, "train_runtime": 42522.1813, "train_samples_per_second": 0.47, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1044519143, "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 }