{ "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": 0.10498046875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1526.0, "completions/mean_length": 461.30107421875, "completions/mean_terminated_length": 335.2909790039063, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.00379525707103312, "learning_rate": 3.1249999999999997e-07, "loss": 0.0469, "num_tokens": 19567755.0, "reward": 0.21255219578742982, "reward_std": 0.3932221710681915, "rewards/accuracy_reward": 0.06689453125, "rewards/brier_reward": 0.11327975392341613, "rewards/confidence_one_or_zero": 0.41611328125, "rewards/format_reward": 0.26865234375, "rewards/mean_confidence_reward": 0.5411806344985962, "rewards/smce": -0.474483048915863, "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": 0.09873046875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1530.6, "completions/mean_length": 437.7197265625, "completions/mean_terminated_length": 317.39723510742186, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.0042224242351949215, "learning_rate": 6.249999999999999e-07, "loss": 0.0463, "num_tokens": 39150357.0, "reward": 0.22614014744758607, "reward_std": 0.39391623735427855, "rewards/accuracy_reward": 0.062109375, "rewards/brier_reward": 0.12040608525276184, "rewards/confidence_one_or_zero": 0.4322265625, "rewards/format_reward": 0.29501953125, "rewards/mean_confidence_reward": 0.5668324828147888, "rewards/smce": -0.505133843421936, "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": 0.05341796875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1478.6, "completions/mean_length": 311.09013671875, "completions/mean_terminated_length": 242.4318878173828, "completions/min_length": 1.4, "completions/min_terminated_length": 1.4, "epoch": 0.048, "grad_norm": 0.00403438089415431, "learning_rate": 9.374999999999999e-07, "loss": 0.0477, "num_tokens": 57384656.0, "reward": 0.38819911479949953, "reward_std": 0.43736083507537843, "rewards/accuracy_reward": 0.10380859375, "rewards/brier_reward": 0.1952836811542511, "rewards/confidence_one_or_zero": 0.53505859375, "rewards/format_reward": 0.5064453125, "rewards/mean_confidence_reward": 0.6857642889022827, "rewards/smce": -0.5828361272811889, "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": 0.02236328125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1449.0, "completions/mean_length": 212.86318359375, "completions/mean_terminated_length": 182.72677917480468, "completions/min_length": 14.2, "completions/min_terminated_length": 14.2, "epoch": 0.064, "grad_norm": 0.0026617355179041624, "learning_rate": 1e-06, "loss": 0.035, "num_tokens": 74482775.0, "reward": 0.6544042944908142, "reward_std": 0.40470498204231264, "rewards/accuracy_reward": 0.18193359375, "rewards/brier_reward": 0.3313737392425537, "rewards/confidence_one_or_zero": 0.56455078125, "rewards/format_reward": 0.82607421875, "rewards/mean_confidence_reward": 0.7916394591331481, "rewards/smce": -0.6115134835243226, "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": 0.0138671875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1197.4, "completions/mean_length": 158.989453125, "completions/mean_terminated_length": 139.62335815429688, "completions/min_length": 17.2, "completions/min_terminated_length": 17.2, "epoch": 0.08, "grad_norm": 0.0025597603525966406, "learning_rate": 1e-06, "loss": 0.0179, "num_tokens": 91043979.0, "reward": 0.7784090042114258, "reward_std": 0.330885636806488, "rewards/accuracy_reward": 0.2107421875, "rewards/brier_reward": 0.4119755744934082, "rewards/confidence_one_or_zero": 0.4875, "rewards/format_reward": 0.96240234375, "rewards/mean_confidence_reward": 0.7723959684371948, "rewards/smce": -0.5660923838615417, "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": 0.00869140625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1161.8, "completions/mean_length": 142.11708984375, "completions/mean_terminated_length": 129.9005920410156, "completions/min_length": 40.4, "completions/min_terminated_length": 40.4, "epoch": 0.096, "grad_norm": 0.002101193880662322, "learning_rate": 1e-06, "loss": 0.014, "num_tokens": 107543866.0, "reward": 0.8724155187606811, "reward_std": 0.2890653073787689, "rewards/accuracy_reward": 0.23212890625, "rewards/brier_reward": 0.5501087605953217, "rewards/confidence_one_or_zero": 0.3072265625, "rewards/format_reward": 0.98203125, "rewards/mean_confidence_reward": 0.603573477268219, "rewards/smce": -0.38879336416721344, "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": 0.00419921875, "completions/max_length": 1536.0, "completions/max_terminated_length": 990.2, "completions/mean_length": 121.371484375, "completions/mean_terminated_length": 115.4102279663086, "completions/min_length": 35.0, "completions/min_terminated_length": 35.0, "epoch": 0.112, "grad_norm": 0.0015070850495249033, "learning_rate": 1e-06, "loss": 0.008, "num_tokens": 123896182.0, "reward": 0.9859124302864075, "reward_std": 0.20140353739261627, "rewards/accuracy_reward": 0.27421875, "rewards/brier_reward": 0.7134735107421875, "rewards/confidence_one_or_zero": 0.2005859375, "rewards/format_reward": 0.99169921875, "rewards/mean_confidence_reward": 0.35798991322517393, "rewards/smce": -0.15135461539030076, "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": 0.0017578125, "completions/max_length": 1361.0, "completions/max_terminated_length": 950.4, "completions/mean_length": 113.27216796875, "completions/mean_terminated_length": 110.7666030883789, "completions/min_length": 40.4, "completions/min_terminated_length": 40.4, "epoch": 0.128, "grad_norm": 0.0008679942111484706, "learning_rate": 1e-06, "loss": 0.0046, "num_tokens": 139972761.0, "reward": 1.0169670820236205, "reward_std": 0.1333804875612259, "rewards/accuracy_reward": 0.2720703125, "rewards/brier_reward": 0.7683740258216858, "rewards/confidence_one_or_zero": 0.15654296875, "rewards/format_reward": 0.9966796875, "rewards/mean_confidence_reward": 0.24390380084514618, "rewards/smce": -0.06381280496716499, "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": 0.00048828125, "completions/max_length": 1156.4, "completions/max_terminated_length": 758.0, "completions/mean_length": 111.1904296875, "completions/mean_terminated_length": 110.49412536621094, "completions/min_length": 49.8, "completions/min_terminated_length": 49.8, "epoch": 0.144, "grad_norm": 0.0008737383177503943, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 156061783.0, "reward": 1.0350018739700317, "reward_std": 0.10654668658971786, "rewards/accuracy_reward": 0.32724609375, "rewards/brier_reward": 0.7511680364608765, "rewards/confidence_one_or_zero": 0.10390625, "rewards/format_reward": 0.99775390625, "rewards/mean_confidence_reward": 0.21236729323863984, "rewards/smce": -0.12329771071672439, "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": 0.00087890625, "completions/max_length": 1536.0, "completions/max_terminated_length": 706.4, "completions/mean_length": 114.46494140625, "completions/mean_terminated_length": 113.21427154541016, "completions/min_length": 44.0, "completions/min_terminated_length": 44.0, "epoch": 0.16, "grad_norm": 0.0007549422443844378, "learning_rate": 1e-06, "loss": 0.0019, "num_tokens": 172254832.0, "reward": 1.0396341800689697, "reward_std": 0.09648548364639283, "rewards/accuracy_reward": 0.321875, "rewards/brier_reward": 0.7635697841644287, "rewards/confidence_one_or_zero": 0.046484375, "rewards/format_reward": 0.9984375, "rewards/mean_confidence_reward": 0.23609647154808044, "rewards/smce": -0.09228837788105011, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.002049180327868827, "eval_completions/max_length": 928.0, "eval_completions/max_terminated_length": 374.0, "eval_completions/mean_length": 117.9098014831543, "eval_completions/mean_terminated_length": 115.00321960449219, "eval_completions/min_length": 56.0, "eval_completions/min_terminated_length": 56.0, "eval_loss": 0.0, "eval_num_tokens": 172254832.0, "eval_reward": 1.0303497314453125, "eval_reward_std": 0.14893124997615814, "eval_rewards/accuracy_reward": 0.26171875, "eval_rewards/brier_reward": 0.8018642067909241, "eval_rewards/confidence_one_or_zero": 0.0234375, "eval_rewards/format_reward": 0.998046875, "eval_rewards/mean_confidence_reward": 0.26405274868011475, "eval_rewards/smce": -0.01861918903887272, "eval_runtime": 21.6493, "eval_samples_per_second": 23.095, "eval_steps_per_second": 0.092, "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": 9.765625e-05, "completions/max_length": 709.8, "completions/max_terminated_length": 462.6, "completions/mean_length": 114.0119140625, "completions/mean_terminated_length": 113.8730453491211, "completions/min_length": 47.6, "completions/min_terminated_length": 47.6, "epoch": 0.176, "grad_norm": 0.0008791896980255842, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 188659434.0, "reward": 1.0433724403381348, "reward_std": 0.09667095988988876, "rewards/accuracy_reward": 0.30830078125, "rewards/brier_reward": 0.781142795085907, "rewards/confidence_one_or_zero": 0.02275390625, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.27407650351524354, "rewards/smce": -0.04226556643843651, "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": 9.765625e-05, "completions/max_length": 635.2, "completions/max_terminated_length": 415.8, "completions/mean_length": 113.58115234375, "completions/mean_terminated_length": 113.4424545288086, "completions/min_length": 38.2, "completions/min_terminated_length": 38.2, "epoch": 0.192, "grad_norm": 0.0017089458415284753, "learning_rate": 1e-06, "loss": -0.0005, "num_tokens": 204637321.0, "reward": 1.0613579511642457, "reward_std": 0.105955308675766, "rewards/accuracy_reward": 0.348828125, "rewards/brier_reward": 0.7770011067390442, "rewards/confidence_one_or_zero": 0.0076171875, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.3225338637828827, "rewards/smce": -0.050558538734912874, "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": 826.0, "completions/max_terminated_length": 335.0, "completions/mean_length": 116.15986328125, "completions/mean_terminated_length": 115.88248596191406, "completions/min_length": 56.4, "completions/min_terminated_length": 56.4, "epoch": 0.208, "grad_norm": 0.0009319775272160769, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 220859022.0, "reward": 1.066517448425293, "reward_std": 0.11409244835376739, "rewards/accuracy_reward": 0.36494140625, "rewards/brier_reward": 0.7699201822280883, "rewards/confidence_one_or_zero": 0.0037109375, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.36000165343284607, "rewards/smce": -0.030688709765672683, "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": 0.00048828125, "completions/max_length": 1148.6, "completions/max_terminated_length": 409.6, "completions/mean_length": 114.28154296875, "completions/mean_terminated_length": 113.58670349121094, "completions/min_length": 55.0, "completions/min_terminated_length": 55.0, "epoch": 0.224, "grad_norm": 0.0010573499603196979, "learning_rate": 1e-06, "loss": 0.002, "num_tokens": 237182449.0, "reward": 1.0665554523468017, "reward_std": 0.1144440695643425, "rewards/accuracy_reward": 0.36416015625, "rewards/brier_reward": 0.7713935971260071, "rewards/confidence_one_or_zero": 0.00244140625, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.3705263793468475, "rewards/smce": -0.03520021215081215, "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": 537.6, "completions/max_terminated_length": 289.6, "completions/mean_length": 112.665234375, "completions/mean_terminated_length": 112.52650146484375, "completions/min_length": 55.4, "completions/min_terminated_length": 55.4, "epoch": 0.24, "grad_norm": 0.0010807751677930355, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 253587821.0, "reward": 1.0801397800445556, "reward_std": 0.1174217015504837, "rewards/accuracy_reward": 0.40361328125, "rewards/brier_reward": 0.7600827097892762, "rewards/confidence_one_or_zero": 0.0009765625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3807018220424652, "rewards/smce": -0.06443720050156117, "step": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 800.0, "completions/max_terminated_length": 363.2, "completions/mean_length": 110.91630859375, "completions/mean_terminated_length": 110.63789825439453, "completions/min_length": 56.8, "completions/min_terminated_length": 56.8, "epoch": 0.256, "grad_norm": 0.0011304247891530395, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 269778420.0, "reward": 1.0721617221832276, "reward_std": 0.11495736688375473, "rewards/accuracy_reward": 0.3783203125, "rewards/brier_reward": 0.7689534902572632, "rewards/confidence_one_or_zero": 0.00068359375, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3930156409740448, "rewards/smce": -0.05120910629630089, "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": 0.0, "completions/max_length": 342.2, "completions/max_terminated_length": 342.2, "completions/mean_length": 112.69873046875, "completions/mean_terminated_length": 112.69873046875, "completions/min_length": 57.2, "completions/min_terminated_length": 57.2, "epoch": 0.272, "grad_norm": 0.0010486873798072338, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 285898151.0, "reward": 1.0783493280410767, "reward_std": 0.1138525053858757, "rewards/accuracy_reward": 0.39580078125, "rewards/brier_reward": 0.7631247043609619, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.39648603796958926, "rewards/smce": -0.03478789925575256, "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": 9.765625e-05, "completions/max_length": 564.8, "completions/max_terminated_length": 317.4, "completions/mean_length": 113.1732421875, "completions/mean_terminated_length": 113.0341781616211, "completions/min_length": 45.2, "completions/min_terminated_length": 45.2, "epoch": 0.288, "grad_norm": 0.0010555179324001074, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 302015221.0, "reward": 1.0807376623153686, "reward_std": 0.10946670323610305, "rewards/accuracy_reward": 0.39296875, "rewards/brier_reward": 0.7711483836174011, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3924079716205597, "rewards/smce": -0.04894471056759357, "step": 90 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 674.2, "completions/max_terminated_length": 423.2, "completions/mean_length": 115.8076171875, "completions/mean_terminated_length": 115.66913299560547, "completions/min_length": 55.0, "completions/min_terminated_length": 55.0, "epoch": 0.304, "grad_norm": 0.0011386368423700333, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 318131043.0, "reward": 1.0865533351898193, "reward_std": 0.10848731994628906, "rewards/accuracy_reward": 0.408203125, "rewards/brier_reward": 0.7670220613479615, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.39132454991340637, "rewards/smce": -0.04043112620711327, "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": 9.765625e-05, "completions/max_length": 572.8, "completions/max_terminated_length": 326.2, "completions/mean_length": 116.87548828125, "completions/mean_terminated_length": 116.7369140625, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.32, "grad_norm": 0.0009122883784584701, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 334416552.0, "reward": 1.0942379474639892, "reward_std": 0.1046961173415184, "rewards/accuracy_reward": 0.425390625, "rewards/brier_reward": 0.7664251208305359, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.3974118232727051, "rewards/smce": -0.06485869474709034, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 282.0, "eval_completions/max_terminated_length": 282.0, "eval_completions/mean_length": 119.43567276000977, "eval_completions/mean_terminated_length": 119.43567276000977, "eval_completions/min_length": 62.0, "eval_completions/min_terminated_length": 62.0, "eval_loss": 0.0, "eval_num_tokens": 334416552.0, "eval_reward": 1.0595381259918213, "eval_reward_std": 0.2181749865412712, "eval_rewards/accuracy_reward": 0.34765625, "eval_rewards/brier_reward": 0.7744396030902863, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4042382836341858, "eval_rewards/smce": -0.060410160571336746, "eval_runtime": 11.1966, "eval_samples_per_second": 44.656, "eval_steps_per_second": 0.179, "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": 598.4, "completions/max_terminated_length": 347.4, "completions/mean_length": 116.05048828125, "completions/mean_terminated_length": 115.91200714111328, "completions/min_length": 56.2, "completions/min_terminated_length": 56.2, "epoch": 0.336, "grad_norm": 0.000933317409362644, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 350327341.0, "reward": 1.0963164806365966, "reward_std": 0.10393176078796387, "rewards/accuracy_reward": 0.42666015625, "rewards/brier_reward": 0.7689025402069092, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.40582754015922545, "rewards/smce": -0.0527496300637722, "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": 0.0001953125, "completions/max_length": 834.4, "completions/max_terminated_length": 328.2, "completions/mean_length": 119.27626953125, "completions/mean_terminated_length": 118.99936370849609, "completions/min_length": 58.2, "completions/min_terminated_length": 58.2, "epoch": 0.352, "grad_norm": 0.0010004864307120442, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 366809146.0, "reward": 1.0733874320983887, "reward_std": 0.1073990985751152, "rewards/accuracy_reward": 0.37646484375, "rewards/brier_reward": 0.7720993638038636, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4040791392326355, "rewards/smce": -0.029942740127444266, "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": 9.765625e-05, "completions/max_length": 610.2, "completions/max_terminated_length": 363.6, "completions/mean_length": 123.4861328125, "completions/mean_terminated_length": 123.34794921875, "completions/min_length": 59.4, "completions/min_terminated_length": 59.4, "epoch": 0.368, "grad_norm": 0.0008533200598321855, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 383139116.0, "reward": 1.0894709587097169, "reward_std": 0.10314081609249115, "rewards/accuracy_reward": 0.4060546875, "rewards/brier_reward": 0.7751178503036499, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4102395951747894, "rewards/smce": -0.038769982382655145, "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": 594.6, "completions/max_terminated_length": 343.8, "completions/mean_length": 126.59150390625, "completions/mean_terminated_length": 126.45359497070312, "completions/min_length": 59.6, "completions/min_terminated_length": 59.6, "epoch": 0.384, "grad_norm": 0.0008750154520384967, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 399291925.0, "reward": 1.1030329465866089, "reward_std": 0.09855032563209534, "rewards/accuracy_reward": 0.434765625, "rewards/brier_reward": 0.7735632658004761, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.411165726184845, "rewards/smce": -0.0433216568082571, "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": 576.6, "completions/max_terminated_length": 332.0, "completions/mean_length": 131.9685546875, "completions/mean_terminated_length": 131.83190002441407, "completions/min_length": 61.8, "completions/min_terminated_length": 61.8, "epoch": 0.4, "grad_norm": 0.0009054332040250301, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 415679731.0, "reward": 1.090169644355774, "reward_std": 0.11003399789333343, "rewards/accuracy_reward": 0.401171875, "rewards/brier_reward": 0.7810108900070191, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.42416959404945376, "rewards/smce": -0.03493335172533989, "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": 0.0005859375, "completions/max_length": 1102.2, "completions/max_terminated_length": 408.4, "completions/mean_length": 136.919921875, "completions/mean_terminated_length": 136.10001220703126, "completions/min_length": 61.0, "completions/min_terminated_length": 61.0, "epoch": 0.416, "grad_norm": 0.0008075532387010753, "learning_rate": 1e-06, "loss": 0.0017, "num_tokens": 431962975.0, "reward": 1.0986106395721436, "reward_std": 0.11043254286050797, "rewards/accuracy_reward": 0.41943359375, "rewards/brier_reward": 0.7794132232666016, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.4245561957359314, "rewards/smce": -0.020808039605617522, "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": 9.765625e-05, "completions/max_length": 576.6, "completions/max_terminated_length": 359.8, "completions/mean_length": 137.3033203125, "completions/mean_terminated_length": 137.16665344238282, "completions/min_length": 62.6, "completions/min_terminated_length": 62.6, "epoch": 0.432, "grad_norm": 0.0010306270560249686, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 448383297.0, "reward": 1.1073107242584228, "reward_std": 0.10405029356479645, "rewards/accuracy_reward": 0.443359375, "rewards/brier_reward": 0.7737538814544678, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4363938868045807, "rewards/smce": -0.04594621378928423, "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": 9.765625e-05, "completions/max_length": 653.8, "completions/max_terminated_length": 442.0, "completions/mean_length": 137.083984375, "completions/mean_terminated_length": 136.94732055664062, "completions/min_length": 61.2, "completions/min_terminated_length": 61.2, "epoch": 0.448, "grad_norm": 0.0007984083495102823, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 464739837.0, "reward": 1.101181983947754, "reward_std": 0.10962242484092713, "rewards/accuracy_reward": 0.43056640625, "rewards/brier_reward": 0.7736897945404053, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4366616249084473, "rewards/smce": -0.0359076976776123, "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": 0.000390625, "completions/max_length": 1080.6, "completions/max_terminated_length": 377.0, "completions/mean_length": 137.136328125, "completions/mean_terminated_length": 136.58931884765624, "completions/min_length": 66.0, "completions/min_terminated_length": 66.0, "epoch": 0.464, "grad_norm": 0.0007966695702634752, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 481314929.0, "reward": 1.079134178161621, "reward_std": 0.10293117165565491, "rewards/accuracy_reward": 0.3755859375, "rewards/brier_reward": 0.7862482190132141, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.4163307726383209, "rewards/smce": -0.06156680956482887, "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": 652.8, "completions/max_terminated_length": 430.8, "completions/mean_length": 134.71728515625, "completions/mean_terminated_length": 134.58053283691407, "completions/min_length": 65.8, "completions/min_terminated_length": 65.8, "epoch": 0.48, "grad_norm": 0.0009112542029470205, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 497742466.0, "reward": 1.1103950500488282, "reward_std": 0.10099990367889404, "rewards/accuracy_reward": 0.4515625, "rewards/brier_reward": 0.772011113166809, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4020026683807373, "rewards/smce": -0.05373247638344765, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 300.5, "eval_completions/max_terminated_length": 300.5, "eval_completions/mean_length": 135.326904296875, "eval_completions/mean_terminated_length": 135.326904296875, "eval_completions/min_length": 74.0, "eval_completions/min_terminated_length": 74.0, "eval_loss": 0.0, "eval_num_tokens": 497742466.0, "eval_reward": 1.067869484424591, "eval_reward_std": 0.22196416556835175, "eval_rewards/accuracy_reward": 0.349609375, "eval_rewards/brier_reward": 0.788908839225769, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.40415431559085846, "eval_rewards/smce": -0.05559943988919258, "eval_runtime": 10.8135, "eval_samples_per_second": 46.239, "eval_steps_per_second": 0.185, "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": 597.0, "completions/max_terminated_length": 354.8, "completions/mean_length": 133.0171875, "completions/mean_terminated_length": 132.8801483154297, "completions/min_length": 53.2, "completions/min_terminated_length": 53.2, "epoch": 0.496, "grad_norm": 0.0009701360249891877, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 514412402.0, "reward": 1.111799955368042, "reward_std": 0.10661143213510513, "rewards/accuracy_reward": 0.4494140625, "rewards/brier_reward": 0.7759701251983643, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.42597988843917844, "rewards/smce": -0.03179615084081888, "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": 9.765625e-05, "completions/max_length": 571.6, "completions/max_terminated_length": 331.2, "completions/mean_length": 130.59658203125, "completions/mean_terminated_length": 130.4593994140625, "completions/min_length": 63.0, "completions/min_terminated_length": 63.0, "epoch": 0.512, "grad_norm": 0.0012009345227852464, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 530895375.0, "reward": 1.1192145109176637, "reward_std": 0.11020909100770951, "rewards/accuracy_reward": 0.459765625, "rewards/brier_reward": 0.7806031942367554, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4468603491783142, "rewards/smce": -0.034905324690043926, "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": 613.2, "completions/max_terminated_length": 359.8, "completions/mean_length": 137.18359375, "completions/mean_terminated_length": 137.04674682617187, "completions/min_length": 65.4, "completions/min_terminated_length": 65.4, "epoch": 0.528, "grad_norm": 0.001150839147157967, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 547329671.0, "reward": 1.1141958951950073, "reward_std": 0.11657649278640747, "rewards/accuracy_reward": 0.44794921875, "rewards/brier_reward": 0.7825037598609924, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.46873701810836793, "rewards/smce": -0.035382162779569626, "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": 611.6, "completions/max_terminated_length": 386.8, "completions/mean_length": 142.52001953125, "completions/mean_terminated_length": 142.38402099609374, "completions/min_length": 65.6, "completions/min_terminated_length": 65.6, "epoch": 0.544, "grad_norm": 0.0009249963914044201, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 563952660.0, "reward": 1.1343823671340942, "reward_std": 0.12186299562454224, "rewards/accuracy_reward": 0.49697265625, "rewards/brier_reward": 0.773317551612854, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.48627060651779175, "rewards/smce": -0.02662253864109516, "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": 480.6, "completions/max_terminated_length": 480.6, "completions/mean_length": 146.33427734375, "completions/mean_terminated_length": 146.33427734375, "completions/min_length": 69.8, "completions/min_terminated_length": 69.8, "epoch": 0.56, "grad_norm": 0.000978289870545268, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 580272531.0, "reward": 1.104417610168457, "reward_std": 0.11959323734045028, "rewards/accuracy_reward": 0.4271484375, "rewards/brier_reward": 0.7850151300430298, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49268749356269836, "rewards/smce": -0.0665838647633791, "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": 0.00048828125, "completions/max_length": 1085.0, "completions/max_terminated_length": 499.2, "completions/mean_length": 155.02001953125, "completions/mean_terminated_length": 154.34617614746094, "completions/min_length": 69.4, "completions/min_terminated_length": 69.4, "epoch": 0.576, "grad_norm": 0.0006955881253816187, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 597046560.0, "reward": 1.10858473777771, "reward_std": 0.10167575627565384, "rewards/accuracy_reward": 0.4353515625, "rewards/brier_reward": 0.7846779823303223, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.4585468828678131, "rewards/smce": -0.04745281338691711, "step": 180 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 582.6, "completions/max_terminated_length": 582.6, "completions/mean_length": 158.10908203125, "completions/mean_terminated_length": 158.10908203125, "completions/min_length": 80.2, "completions/min_terminated_length": 80.2, "epoch": 0.592, "grad_norm": 0.0007171398028731346, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 613833309.0, "reward": 1.1120543956756592, "reward_std": 0.091026870906353, "rewards/accuracy_reward": 0.44326171875, "rewards/brier_reward": 0.7823067307472229, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.42848995327949524, "rewards/smce": -0.029207753576338293, "step": 185 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 636.6, "completions/max_terminated_length": 445.0, "completions/mean_length": 161.95458984375, "completions/mean_terminated_length": 161.6872802734375, "completions/min_length": 80.0, "completions/min_terminated_length": 80.0, "epoch": 0.608, "grad_norm": 0.0007596832583658397, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 630491212.0, "reward": 1.117729663848877, "reward_std": 0.09016477465629577, "rewards/accuracy_reward": 0.44638671875, "rewards/brier_reward": 0.7919416666030884, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.40303613543510436, "rewards/smce": -0.053490759804844856, "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": 806.8, "completions/max_terminated_length": 591.6, "completions/mean_length": 168.21943359375, "completions/mean_terminated_length": 168.085693359375, "completions/min_length": 75.0, "completions/min_terminated_length": 75.0, "epoch": 0.624, "grad_norm": 0.000917185447178781, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 647557683.0, "reward": 1.1099303245544434, "reward_std": 0.09768695533275604, "rewards/accuracy_reward": 0.43203125, "rewards/brier_reward": 0.7900554180145264, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.42108300924301145, "rewards/smce": -0.04258261527866125, "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": 844.0, "completions/max_terminated_length": 698.8, "completions/mean_length": 168.75458984375, "completions/mean_terminated_length": 168.6212585449219, "completions/min_length": 78.8, "completions/min_terminated_length": 78.8, "epoch": 0.64, "grad_norm": 0.000845664064399898, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 664628418.0, "reward": 1.148032546043396, "reward_std": 0.10199344009160996, "rewards/accuracy_reward": 0.50791015625, "rewards/brier_reward": 0.7912523984909058, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4784619212150574, "rewards/smce": -0.06001344136893749, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 344.0, "eval_completions/max_terminated_length": 344.0, "eval_completions/mean_length": 172.08222198486328, "eval_completions/mean_terminated_length": 172.08222198486328, "eval_completions/min_length": 93.0, "eval_completions/min_terminated_length": 93.0, "eval_loss": 0.0, "eval_num_tokens": 664628418.0, "eval_reward": 1.0765975713729858, "eval_reward_std": 0.27346640825271606, "eval_rewards/accuracy_reward": 0.375, "eval_rewards/brier_reward": 0.7840966582298279, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4926757961511612, "eval_rewards/smce": -0.11805175989866257, "eval_runtime": 11.8017, "eval_samples_per_second": 42.367, "eval_steps_per_second": 0.169, "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": 0.0001953125, "completions/max_length": 937.0, "completions/max_terminated_length": 524.4, "completions/mean_length": 166.0533203125, "completions/mean_terminated_length": 165.78526916503907, "completions/min_length": 77.0, "completions/min_terminated_length": 77.0, "epoch": 0.656, "grad_norm": 0.0011595977703109384, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 681185348.0, "reward": 1.095522141456604, "reward_std": 0.1261553093791008, "rewards/accuracy_reward": 0.41953125, "rewards/brier_reward": 0.7763510584831238, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5122119128704071, "rewards/smce": -0.09287466555833816, "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": 926.4, "completions/max_terminated_length": 485.8, "completions/mean_length": 161.9060546875, "completions/mean_terminated_length": 161.50301208496094, "completions/min_length": 75.0, "completions/min_terminated_length": 75.0, "epoch": 0.672, "grad_norm": 0.0007186342845670879, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 697756706.0, "reward": 1.1261529922485352, "reward_std": 0.09019538462162018, "rewards/accuracy_reward": 0.456640625, "rewards/brier_reward": 0.7979784965515136, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4447070300579071, "rewards/smce": -0.040419237315654756, "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": 9.765625e-05, "completions/max_length": 696.0, "completions/max_terminated_length": 456.6, "completions/mean_length": 159.845703125, "completions/mean_terminated_length": 159.7115234375, "completions/min_length": 78.4, "completions/min_terminated_length": 78.4, "epoch": 0.688, "grad_norm": 0.0008274798165075481, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 714347446.0, "reward": 1.1239240884780883, "reward_std": 0.08857346624135971, "rewards/accuracy_reward": 0.45234375, "rewards/brier_reward": 0.7981483459472656, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4173779249191284, "rewards/smce": -0.05094240251928568, "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": 460.4, "completions/max_terminated_length": 460.4, "completions/mean_length": 161.32236328125, "completions/mean_terminated_length": 161.32236328125, "completions/min_length": 78.4, "completions/min_terminated_length": 78.4, "epoch": 0.704, "grad_norm": 0.0009050744120031595, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 730865531.0, "reward": 1.1346723556518554, "reward_std": 0.0825655698776245, "rewards/accuracy_reward": 0.48037109375, "rewards/brier_reward": 0.7921864032745362, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.41924610137939455, "rewards/smce": -0.06231907419860363, "step": 220 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 635.0, "completions/max_terminated_length": 418.2, "completions/mean_length": 161.794921875, "completions/mean_terminated_length": 161.6607177734375, "completions/min_length": 77.8, "completions/min_terminated_length": 77.8, "epoch": 0.72, "grad_norm": 0.0010375964920967817, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 747532167.0, "reward": 1.1471279382705688, "reward_std": 0.09896319657564163, "rewards/accuracy_reward": 0.48896484375, "rewards/brier_reward": 0.8072811841964722, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.48550487756729127, "rewards/smce": -0.03786895275115967, "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": 9.765625e-05, "completions/max_length": 639.8, "completions/max_terminated_length": 414.8, "completions/mean_length": 162.24970703125, "completions/mean_terminated_length": 162.1154052734375, "completions/min_length": 80.2, "completions/min_terminated_length": 80.2, "epoch": 0.736, "grad_norm": 0.0011135365348309278, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 764133188.0, "reward": 1.1404980421066284, "reward_std": 0.11243823915719986, "rewards/accuracy_reward": 0.49609375, "rewards/brier_reward": 0.7878056287765502, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5388007879257202, "rewards/smce": -0.056135011836886406, "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": 422.2, "completions/max_terminated_length": 422.2, "completions/mean_length": 163.6078125, "completions/mean_terminated_length": 163.6078125, "completions/min_length": 75.4, "completions/min_terminated_length": 75.4, "epoch": 0.752, "grad_norm": 0.0010154708288609982, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 781035732.0, "reward": 1.1349368810653686, "reward_std": 0.1062909796833992, "rewards/accuracy_reward": 0.47587890625, "rewards/brier_reward": 0.7961811423301697, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4935058653354645, "rewards/smce": -0.04374487325549126, "step": 235 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 416.6, "completions/max_terminated_length": 416.6, "completions/mean_length": 169.75712890625, "completions/mean_terminated_length": 169.75712890625, "completions/min_length": 79.6, "completions/min_terminated_length": 79.6, "epoch": 0.768, "grad_norm": 0.0009220530628226697, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 797706749.0, "reward": 1.1410370349884034, "reward_std": 0.09544525891542435, "rewards/accuracy_reward": 0.47705078125, "rewards/brier_reward": 0.8060692191123963, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4754668116569519, "rewards/smce": -0.018984863162040712, "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": 428.4, "completions/max_terminated_length": 428.4, "completions/mean_length": 171.64931640625, "completions/mean_terminated_length": 171.64931640625, "completions/min_length": 84.4, "completions/min_terminated_length": 84.4, "epoch": 0.784, "grad_norm": 0.0008822088711895049, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 814638806.0, "reward": 1.1337279319763183, "reward_std": 0.10449115782976151, "rewards/accuracy_reward": 0.4818359375, "rewards/brier_reward": 0.7891198754310608, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.48597265481948854, "rewards/smce": -0.0700188372284174, "step": 245 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1085.4, "completions/max_terminated_length": 453.6, "completions/mean_length": 170.1076171875, "completions/mean_terminated_length": 169.7075622558594, "completions/min_length": 85.2, "completions/min_terminated_length": 85.2, "epoch": 0.8, "grad_norm": 0.0010923626832664013, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 831391268.0, "reward": 1.151297688484192, "reward_std": 0.11599037945270538, "rewards/accuracy_reward": 0.5, "rewards/brier_reward": 0.805100429058075, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.523035180568695, "rewards/smce": -0.04035660531371832, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 318.0, "eval_completions/max_terminated_length": 318.0, "eval_completions/mean_length": 170.98802185058594, "eval_completions/mean_terminated_length": 170.98802185058594, "eval_completions/min_length": 88.5, "eval_completions/min_terminated_length": 88.5, "eval_loss": 0.0, "eval_num_tokens": 831391268.0, "eval_reward": 1.0831419825553894, "eval_reward_std": 0.28824666142463684, "eval_rewards/accuracy_reward": 0.375, "eval_rewards/brier_reward": 0.7978320419788361, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.5054687708616257, "eval_rewards/smce": -0.1309814490377903, "eval_runtime": 11.3528, "eval_samples_per_second": 44.042, "eval_steps_per_second": 0.176, "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": 0.00029296875, "completions/max_length": 1094.8, "completions/max_terminated_length": 434.6, "completions/mean_length": 169.755078125, "completions/mean_terminated_length": 169.3546356201172, "completions/min_length": 83.6, "completions/min_terminated_length": 83.6, "epoch": 0.816, "grad_norm": 0.0011063864221796393, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 848228728.0, "reward": 1.1495605945587157, "reward_std": 0.11100535690784455, "rewards/accuracy_reward": 0.49404296875, "rewards/brier_reward": 0.806819224357605, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5094443440437317, "rewards/smce": -0.02897880058735609, "step": 255 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 651.6, "completions/max_terminated_length": 435.2, "completions/mean_length": 176.95146484375, "completions/mean_terminated_length": 176.8187255859375, "completions/min_length": 87.4, "completions/min_terminated_length": 87.4, "epoch": 0.832, "grad_norm": 0.001013425295241177, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 865049063.0, "reward": 1.1530806064605712, "reward_std": 0.09840368032455445, "rewards/accuracy_reward": 0.4978515625, "rewards/brier_reward": 0.8105565905570984, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.463302743434906, "rewards/smce": -0.04300446677953005, "step": 260 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 676.0, "completions/max_terminated_length": 439.2, "completions/mean_length": 172.416015625, "completions/mean_terminated_length": 172.28269958496094, "completions/min_length": 86.4, "completions/min_terminated_length": 86.4, "epoch": 0.848, "grad_norm": 0.0009189588017761707, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 881828971.0, "reward": 1.1401946306228639, "reward_std": 0.11134238541126251, "rewards/accuracy_reward": 0.47490234375, "rewards/brier_reward": 0.8072995543479919, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4904062569141388, "rewards/smce": -0.034317873790860175, "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": 0.000390625, "completions/max_length": 881.4, "completions/max_terminated_length": 436.8, "completions/mean_length": 174.81015625, "completions/mean_terminated_length": 174.2799499511719, "completions/min_length": 81.6, "completions/min_terminated_length": 81.6, "epoch": 0.864, "grad_norm": 0.0008723426144570112, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 898605843.0, "reward": 1.1487454652786255, "reward_std": 0.10216106027364731, "rewards/accuracy_reward": 0.5072265625, "rewards/brier_reward": 0.79281005859375, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.49146581888198854, "rewards/smce": -0.04311874024569988, "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": 9.765625e-05, "completions/max_length": 694.2, "completions/max_terminated_length": 498.8, "completions/mean_length": 179.809375, "completions/mean_terminated_length": 179.677294921875, "completions/min_length": 93.6, "completions/min_terminated_length": 93.6, "epoch": 0.88, "grad_norm": 0.0009168738033622503, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 915594163.0, "reward": 1.1366500377655029, "reward_std": 0.08010695576667785, "rewards/accuracy_reward": 0.462890625, "rewards/brier_reward": 0.8138427495956421, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.41804786324501036, "rewards/smce": -0.0667283084243536, "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": 736.6, "completions/max_terminated_length": 637.4, "completions/mean_length": 185.1380859375, "completions/mean_terminated_length": 185.00626220703126, "completions/min_length": 94.0, "completions/min_terminated_length": 94.0, "epoch": 0.896, "grad_norm": 0.0009079832234419882, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 932600825.0, "reward": 1.1398839712142945, "reward_std": 0.09819626659154893, "rewards/accuracy_reward": 0.47734375, "rewards/brier_reward": 0.8047329187393188, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47663772106170654, "rewards/smce": -0.04424012675881386, "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": 0.0001953125, "completions/max_length": 973.0, "completions/max_terminated_length": 515.2, "completions/mean_length": 186.6216796875, "completions/mean_terminated_length": 186.35760803222655, "completions/min_length": 93.6, "completions/min_terminated_length": 93.6, "epoch": 0.912, "grad_norm": 0.0010032489662989974, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 949563127.0, "reward": 1.1336450338363648, "reward_std": 0.11059280037879944, "rewards/accuracy_reward": 0.47021484375, "rewards/brier_reward": 0.8010426044464112, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5327714800834655, "rewards/smce": -0.07546240240335464, "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": 521.8, "completions/max_terminated_length": 521.8, "completions/mean_length": 178.80126953125, "completions/mean_terminated_length": 178.80126953125, "completions/min_length": 94.2, "completions/min_terminated_length": 94.2, "epoch": 0.928, "grad_norm": 0.0009838527766987681, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 966420868.0, "reward": 1.1457917451858521, "reward_std": 0.09142312407493591, "rewards/accuracy_reward": 0.4966796875, "rewards/brier_reward": 0.7967879176139832, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49010254740715026, "rewards/smce": -0.037700850144028665, "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": 630.2, "completions/max_terminated_length": 428.4, "completions/mean_length": 169.31669921875, "completions/mean_terminated_length": 169.18387145996093, "completions/min_length": 86.0, "completions/min_terminated_length": 86.0, "epoch": 0.944, "grad_norm": 0.000964009203016758, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 983130095.0, "reward": 1.1439347505569457, "reward_std": 0.09324800819158555, "rewards/accuracy_reward": 0.494921875, "rewards/brier_reward": 0.7968281626701355, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4685791075229645, "rewards/smce": -0.07567493021488189, "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": 9.765625e-05, "completions/max_length": 664.6, "completions/max_terminated_length": 436.0, "completions/mean_length": 163.23974609375, "completions/mean_terminated_length": 163.10575256347656, "completions/min_length": 83.4, "completions/min_terminated_length": 83.4, "epoch": 0.96, "grad_norm": 0.001006408827379346, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 999741990.0, "reward": 1.1354127168655395, "reward_std": 0.08625964224338531, "rewards/accuracy_reward": 0.4607421875, "rewards/brier_reward": 0.8117612481117249, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4783691465854645, "rewards/smce": -0.03162573650479317, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 319.0, "eval_completions/max_terminated_length": 319.0, "eval_completions/mean_length": 158.7119598388672, "eval_completions/mean_terminated_length": 158.7119598388672, "eval_completions/min_length": 101.0, "eval_completions/min_terminated_length": 101.0, "eval_loss": 0.0, "eval_num_tokens": 999741990.0, "eval_reward": 1.0915340185165405, "eval_reward_std": 0.27021801471710205, "eval_rewards/accuracy_reward": 0.384765625, "eval_rewards/brier_reward": 0.8024950921535492, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4678710997104645, "eval_rewards/smce": -0.08387207612395287, "eval_runtime": 11.1096, "eval_samples_per_second": 45.006, "eval_steps_per_second": 0.18, "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": 0.0001953125, "completions/max_length": 877.2, "completions/max_terminated_length": 459.0, "completions/mean_length": 156.80107421875, "completions/mean_terminated_length": 156.53237609863282, "completions/min_length": 80.6, "completions/min_terminated_length": 80.6, "epoch": 0.976, "grad_norm": 0.000879038474522531, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 1016208753.0, "reward": 1.1549882173538208, "reward_std": 0.08301678597927094, "rewards/accuracy_reward": 0.49765625, "rewards/brier_reward": 0.8142663717269898, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4712851524353027, "rewards/smce": -0.03503466323018074, "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": 0.0001953125, "completions/max_length": 880.0, "completions/max_terminated_length": 482.8, "completions/mean_length": 152.35126953125, "completions/mean_terminated_length": 152.08108215332032, "completions/min_length": 75.8, "completions/min_terminated_length": 75.8, "epoch": 0.992, "grad_norm": 0.0010177699150517583, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 1032897310.0, "reward": 1.140252137184143, "reward_std": 0.0865005910396576, "rewards/accuracy_reward": 0.48935546875, "rewards/brier_reward": 0.7936301112174988, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.44715137481689454, "rewards/smce": -0.043783336877822876, "step": 310 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 382.5, "completions/max_terminated_length": 382.5, "completions/mean_length": 155.24308013916016, "completions/mean_terminated_length": 155.24308013916016, "completions/min_length": 78.5, "completions/min_terminated_length": 78.5, "epoch": 0.9984, "num_tokens": 1039540391.0, "reward": 1.1383739113807678, "reward_std": 0.09339407458901405, "rewards/accuracy_reward": 0.4765625, "rewards/brier_reward": 0.8023821711540222, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49127197265625, "rewards/smce": -0.043958742171525955, "step": 312, "total_flos": 0.0, "train_loss": 0.003989085646259273, "train_runtime": 42377.2177, "train_samples_per_second": 0.472, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1039540391, "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 }