{ "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": 0.1080078125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1530.4, "completions/mean_length": 468.44228515625, "completions/mean_terminated_length": 339.2684265136719, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.003270028391852975, "learning_rate": 3.1249999999999997e-07, "loss": 0.0516, "num_tokens": 19640881.0, "reward": 0.22072359919548035, "reward_std": 0.3922470033168793, "rewards/accuracy_reward": 0.065625, "rewards/brier_reward": 0.11248895078897476, "rewards/confidence_one_or_zero": 0.4205078125, "rewards/format_reward": 0.26337890625, "rewards/mean_confidence_reward": 0.540982985496521, "rewards/smce": -0.4757453978061676, "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": 0.10869140625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1529.0, "completions/mean_length": 454.3373046875, "completions/mean_terminated_length": 322.4891845703125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.004311034921556711, "learning_rate": 6.249999999999999e-07, "loss": 0.0536, "num_tokens": 39393647.0, "reward": 0.2296316921710968, "reward_std": 0.38993753790855407, "rewards/accuracy_reward": 0.0630859375, "rewards/brier_reward": 0.11341342478990554, "rewards/confidence_one_or_zero": 0.4353515625, "rewards/format_reward": 0.2828125, "rewards/mean_confidence_reward": 0.5675621032714844, "rewards/smce": -0.5047040998935699, "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": 0.05458984375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1483.2, "completions/mean_length": 319.65009765625, "completions/mean_terminated_length": 250.00528259277343, "completions/min_length": 1.4, "completions/min_terminated_length": 1.4, "epoch": 0.048, "grad_norm": 0.004322660155594349, "learning_rate": 9.374999999999999e-07, "loss": 0.0551, "num_tokens": 57715600.0, "reward": 0.40282471776008605, "reward_std": 0.4364392518997192, "rewards/accuracy_reward": 0.10986328125, "rewards/brier_reward": 0.19759979844093323, "rewards/confidence_one_or_zero": 0.5287109375, "rewards/format_reward": 0.4982421875, "rewards/mean_confidence_reward": 0.6919785380363465, "rewards/smce": -0.582571291923523, "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": 0.022265625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1380.8, "completions/mean_length": 212.587109375, "completions/mean_terminated_length": 182.60447998046874, "completions/min_length": 10.6, "completions/min_terminated_length": 10.6, "epoch": 0.064, "grad_norm": 0.0025710673071444035, "learning_rate": 1e-06, "loss": 0.0366, "num_tokens": 74810892.0, "reward": 0.6773836731910705, "reward_std": 0.3933659136295319, "rewards/accuracy_reward": 0.18837890625, "rewards/brier_reward": 0.3333409786224365, "rewards/confidence_one_or_zero": 0.5728515625, "rewards/format_reward": 0.83310546875, "rewards/mean_confidence_reward": 0.7932586193084716, "rewards/smce": -0.6074068307876587, "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": 0.011328125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1245.4, "completions/mean_length": 162.02568359375, "completions/mean_terminated_length": 146.2960693359375, "completions/min_length": 30.2, "completions/min_terminated_length": 30.2, "epoch": 0.08, "grad_norm": 0.002596383448690176, "learning_rate": 1e-06, "loss": 0.0178, "num_tokens": 91403187.0, "reward": 0.7920157313346863, "reward_std": 0.32526588439941406, "rewards/accuracy_reward": 0.2154296875, "rewards/brier_reward": 0.40586367845535276, "rewards/confidence_one_or_zero": 0.5203125, "rewards/format_reward": 0.96279296875, "rewards/mean_confidence_reward": 0.7853312015533447, "rewards/smce": -0.5747557759284974, "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": 1347.0, "completions/mean_length": 149.4373046875, "completions/mean_terminated_length": 137.28244323730468, "completions/min_length": 28.8, "completions/min_terminated_length": 28.8, "epoch": 0.096, "grad_norm": 0.0020388758275657892, "learning_rate": 1e-06, "loss": 0.0135, "num_tokens": 107978033.0, "reward": 0.8774016618728637, "reward_std": 0.2836435556411743, "rewards/accuracy_reward": 0.23779296875, "rewards/brier_reward": 0.5336522459983826, "rewards/confidence_one_or_zero": 0.34033203125, "rewards/format_reward": 0.9833984375, "rewards/mean_confidence_reward": 0.6492803931236267, "rewards/smce": -0.4229824423789978, "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": 0.0033203125, "completions/max_length": 1536.0, "completions/max_terminated_length": 892.0, "completions/mean_length": 135.47197265625, "completions/mean_terminated_length": 130.8082702636719, "completions/min_length": 41.8, "completions/min_terminated_length": 41.8, "epoch": 0.112, "grad_norm": 0.001400059787556529, "learning_rate": 1e-06, "loss": 0.0061, "num_tokens": 124474738.0, "reward": 0.9876136898994445, "reward_std": 0.21337922811508178, "rewards/accuracy_reward": 0.280859375, "rewards/brier_reward": 0.7013182640075684, "rewards/confidence_one_or_zero": 0.19404296875, "rewards/format_reward": 0.99306640625, "rewards/mean_confidence_reward": 0.4260053992271423, "rewards/smce": -0.17858310043811798, "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": 0.00185546875, "completions/max_length": 1536.0, "completions/max_terminated_length": 760.8, "completions/mean_length": 130.4890625, "completions/mean_terminated_length": 127.876953125, "completions/min_length": 42.8, "completions/min_terminated_length": 42.8, "epoch": 0.128, "grad_norm": 0.0009865883039310575, "learning_rate": 1e-06, "loss": 0.004, "num_tokens": 140727618.0, "reward": 1.0204854726791381, "reward_std": 0.14821541607379912, "rewards/accuracy_reward": 0.2814453125, "rewards/brier_reward": 0.7634376049041748, "rewards/confidence_one_or_zero": 0.13564453125, "rewards/format_reward": 0.99609375, "rewards/mean_confidence_reward": 0.3013033330440521, "rewards/smce": -0.06513335853815079, "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": 0.00068359375, "completions/max_length": 1376.6, "completions/max_terminated_length": 551.0, "completions/mean_length": 126.53388671875, "completions/mean_terminated_length": 125.56988830566407, "completions/min_length": 45.6, "completions/min_terminated_length": 45.6, "epoch": 0.144, "grad_norm": 0.0008671495015732944, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 156973757.0, "reward": 1.0439672470092773, "reward_std": 0.12052493244409561, "rewards/accuracy_reward": 0.32958984375, "rewards/brier_reward": 0.7605997204780579, "rewards/confidence_one_or_zero": 0.10439453125, "rewards/format_reward": 0.99775390625, "rewards/mean_confidence_reward": 0.2419735610485077, "rewards/smce": -0.09777832627296448, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00068359375, "completions/max_length": 1223.6, "completions/max_terminated_length": 700.8, "completions/mean_length": 128.12744140625, "completions/mean_terminated_length": 127.1651123046875, "completions/min_length": 45.8, "completions/min_terminated_length": 45.8, "epoch": 0.16, "grad_norm": 0.0007608377491123974, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 173306710.0, "reward": 1.0441123008728028, "reward_std": 0.10170394033193589, "rewards/accuracy_reward": 0.322265625, "rewards/brier_reward": 0.7680190086364747, "rewards/confidence_one_or_zero": 0.06083984375, "rewards/format_reward": 0.99794921875, "rewards/mean_confidence_reward": 0.22877179086208344, "rewards/smce": -0.09797133281826972, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 296.5, "eval_completions/max_terminated_length": 296.5, "eval_completions/mean_length": 129.6832733154297, "eval_completions/mean_terminated_length": 129.6832733154297, "eval_completions/min_length": 69.0, "eval_completions/min_terminated_length": 69.0, "eval_loss": 0.0, "eval_num_tokens": 173306710.0, "eval_reward": 1.0249468684196472, "eval_reward_std": 0.14161328971385956, "eval_rewards/accuracy_reward": 0.251953125, "eval_rewards/brier_reward": 0.8018492162227631, "eval_rewards/confidence_one_or_zero": 0.041015625, "eval_rewards/format_reward": 0.99609375, "eval_rewards/mean_confidence_reward": 0.24041016399860382, "eval_rewards/smce": -0.02927441243082285, "eval_runtime": 10.8431, "eval_samples_per_second": 46.112, "eval_steps_per_second": 0.184, "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": 1195.4, "completions/max_terminated_length": 487.8, "completions/mean_length": 129.540625, "completions/mean_terminated_length": 129.12831268310546, "completions/min_length": 55.6, "completions/min_terminated_length": 55.6, "epoch": 0.176, "grad_norm": 0.0007807699730619788, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 189870326.0, "reward": 1.0457116842269898, "reward_std": 0.09513775557279587, "rewards/accuracy_reward": 0.3138671875, "rewards/brier_reward": 0.778539526462555, "rewards/confidence_one_or_zero": 0.03935546875, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.2438169986009598, "rewards/smce": -0.07360489293932915, "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": 0.00048828125, "completions/max_length": 1074.6, "completions/max_terminated_length": 433.8, "completions/mean_length": 131.770703125, "completions/mean_terminated_length": 131.08482055664064, "completions/min_length": 59.8, "completions/min_terminated_length": 59.8, "epoch": 0.192, "grad_norm": 0.0007883945945650339, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 206034474.0, "reward": 1.0633267641067505, "reward_std": 0.1018825501203537, "rewards/accuracy_reward": 0.3580078125, "rewards/brier_reward": 0.7696295857429505, "rewards/confidence_one_or_zero": 0.0146484375, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.28199089169502256, "rewards/smce": -0.07945134602487088, "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": 837.0, "completions/max_terminated_length": 386.0, "completions/mean_length": 135.3244140625, "completions/mean_terminated_length": 135.05081176757812, "completions/min_length": 59.8, "completions/min_terminated_length": 59.8, "epoch": 0.208, "grad_norm": 0.0007650046609342098, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 222452420.0, "reward": 1.0677101373672486, "reward_std": 0.10990563184022903, "rewards/accuracy_reward": 0.36865234375, "rewards/brier_reward": 0.7672603011131287, "rewards/confidence_one_or_zero": 0.00673828125, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.32444043159484864, "rewards/smce": -0.0480029359459877, "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": 1536.0, "completions/max_terminated_length": 369.8, "completions/mean_length": 136.843359375, "completions/mean_terminated_length": 136.15984497070312, "completions/min_length": 59.2, "completions/min_terminated_length": 59.2, "epoch": 0.224, "grad_norm": 0.000912521849386394, "learning_rate": 1e-06, "loss": 0.0018, "num_tokens": 239006880.0, "reward": 1.0661633729934692, "reward_std": 0.112481489777565, "rewards/accuracy_reward": 0.36142578125, "rewards/brier_reward": 0.7715874671936035, "rewards/confidence_one_or_zero": 0.00546875, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.3463192105293274, "rewards/smce": -0.03602232336997986, "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": 693.8, "completions/max_terminated_length": 523.8, "completions/mean_length": 140.01494140625, "completions/mean_terminated_length": 139.87867736816406, "completions/min_length": 57.2, "completions/min_terminated_length": 57.2, "epoch": 0.24, "grad_norm": 0.0008707091328687966, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 255692313.0, "reward": 1.0846214294433594, "reward_std": 0.12235956788063049, "rewards/accuracy_reward": 0.4078125, "rewards/brier_reward": 0.7618267297744751, "rewards/confidence_one_or_zero": 0.0013671875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3821188509464264, "rewards/smce": -0.06492935791611672, "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": 819.8, "completions/max_terminated_length": 359.6, "completions/mean_length": 139.777734375, "completions/mean_terminated_length": 139.50499877929687, "completions/min_length": 62.6, "completions/min_terminated_length": 62.6, "epoch": 0.256, "grad_norm": 0.0009150123805738986, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 272178453.0, "reward": 1.072606611251831, "reward_std": 0.12022970616817474, "rewards/accuracy_reward": 0.3763671875, "rewards/brier_reward": 0.7690465927124024, "rewards/confidence_one_or_zero": 0.0009765625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4023988604545593, "rewards/smce": -0.060530804097652435, "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": 0.0001953125, "completions/max_length": 860.4, "completions/max_terminated_length": 387.0, "completions/mean_length": 140.90380859375, "completions/mean_terminated_length": 140.63138427734376, "completions/min_length": 65.6, "completions/min_terminated_length": 65.6, "epoch": 0.272, "grad_norm": 0.0008105236338451505, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 288587004.0, "reward": 1.081153893470764, "reward_std": 0.11917079389095306, "rewards/accuracy_reward": 0.3962890625, "rewards/brier_reward": 0.7664119839668274, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.40586323142051695, "rewards/smce": -0.03432232886552811, "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": 1096.0, "completions/max_terminated_length": 403.2, "completions/mean_length": 137.1322265625, "completions/mean_terminated_length": 136.72223510742188, "completions/min_length": 60.8, "completions/min_terminated_length": 60.8, "epoch": 0.288, "grad_norm": 0.0011007882421836257, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 304949414.0, "reward": 1.0783969640731812, "reward_std": 0.11819728463888168, "rewards/accuracy_reward": 0.39013671875, "rewards/brier_reward": 0.7671497702598572, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.40524903535842893, "rewards/smce": -0.05063976794481277, "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": 607.2, "completions/max_terminated_length": 376.0, "completions/mean_length": 134.31494140625, "completions/mean_terminated_length": 134.17804565429688, "completions/min_length": 59.4, "completions/min_terminated_length": 59.4, "epoch": 0.304, "grad_norm": 0.000939876539632678, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 321254751.0, "reward": 1.084977626800537, "reward_std": 0.11600391864776612, "rewards/accuracy_reward": 0.4052734375, "rewards/brier_reward": 0.7649775862693786, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.389298677444458, "rewards/smce": -0.03614676035940647, "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": 812.2, "completions/max_terminated_length": 347.0, "completions/mean_length": 136.36416015625, "completions/mean_terminated_length": 136.0903289794922, "completions/min_length": 64.6, "completions/min_terminated_length": 64.6, "epoch": 0.32, "grad_norm": 0.000710589752998203, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 337739824.0, "reward": 1.0932400226593018, "reward_std": 0.10608242303133011, "rewards/accuracy_reward": 0.417578125, "rewards/brier_reward": 0.7692002654075623, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3869974732398987, "rewards/smce": -0.0614659234881401, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 300.0, "eval_completions/max_terminated_length": 300.0, "eval_completions/mean_length": 134.9712142944336, "eval_completions/mean_terminated_length": 134.9712142944336, "eval_completions/min_length": 65.5, "eval_completions/min_terminated_length": 65.5, "eval_loss": 0.0, "eval_num_tokens": 337739824.0, "eval_reward": 1.048181176185608, "eval_reward_std": 0.19223156571388245, "eval_rewards/accuracy_reward": 0.306640625, "eval_rewards/brier_reward": 0.7897287011146545, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.3824804872274399, "eval_rewards/smce": -0.07643750682473183, "eval_runtime": 10.9079, "eval_samples_per_second": 45.838, "eval_steps_per_second": 0.183, "step": 100 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 834.0, "completions/max_terminated_length": 349.4, "completions/mean_length": 134.5306640625, "completions/mean_terminated_length": 134.1192596435547, "completions/min_length": 62.2, "completions/min_terminated_length": 62.2, "epoch": 0.336, "grad_norm": 0.0008598133572377264, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 353839850.0, "reward": 1.0955848693847656, "reward_std": 0.10709079802036285, "rewards/accuracy_reward": 0.42373046875, "rewards/brier_reward": 0.7679327011108399, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.3963977575302124, "rewards/smce": -0.05861892774701118, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 348.2, "completions/max_terminated_length": 348.2, "completions/mean_length": 139.5755859375, "completions/mean_terminated_length": 139.5755859375, "completions/min_length": 65.8, "completions/min_terminated_length": 65.8, "epoch": 0.352, "grad_norm": 0.0007286344189196825, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 370529520.0, "reward": 1.075069546699524, "reward_std": 0.10592829287052155, "rewards/accuracy_reward": 0.3759765625, "rewards/brier_reward": 0.7741642475128174, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.3979013681411743, "rewards/smce": -0.024860015138983727, "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": 442.2, "completions/max_terminated_length": 442.2, "completions/mean_length": 146.44921875, "completions/mean_terminated_length": 146.44921875, "completions/min_length": 64.8, "completions/min_terminated_length": 64.8, "epoch": 0.368, "grad_norm": 0.0022383604664355516, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 387094632.0, "reward": 1.088090991973877, "reward_std": 0.11070768386125565, "rewards/accuracy_reward": 0.40341796875, "rewards/brier_reward": 0.7728642702102662, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.41914639472961424, "rewards/smce": -0.034958066418766975, "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": 810.0, "completions/max_terminated_length": 336.2, "completions/mean_length": 154.13203125, "completions/mean_terminated_length": 153.8613250732422, "completions/min_length": 68.0, "completions/min_terminated_length": 68.0, "epoch": 0.384, "grad_norm": 0.0008120255661197007, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 403529456.0, "reward": 1.0988135099411012, "reward_std": 0.10813373029232025, "rewards/accuracy_reward": 0.42109375, "rewards/brier_reward": 0.7767310023307801, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4109645485877991, "rewards/smce": -0.03251770362257957, "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": 370.6, "completions/max_terminated_length": 370.6, "completions/mean_length": 159.397265625, "completions/mean_terminated_length": 159.397265625, "completions/min_length": 68.2, "completions/min_terminated_length": 68.2, "epoch": 0.4, "grad_norm": 0.0007968509453348815, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 420198132.0, "reward": 1.0878613233566283, "reward_std": 0.11208915263414383, "rewards/accuracy_reward": 0.3951171875, "rewards/brier_reward": 0.7806079149246216, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.41420117020606995, "rewards/smce": -0.03192929774522781, "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": 0.0001953125, "completions/max_length": 841.8, "completions/max_terminated_length": 416.2, "completions/mean_length": 164.4671875, "completions/mean_terminated_length": 164.1993438720703, "completions/min_length": 74.8, "completions/min_terminated_length": 74.8, "epoch": 0.416, "grad_norm": 0.0008458816446363926, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 436763460.0, "reward": 1.0922947883605958, "reward_std": 0.11342525780200959, "rewards/accuracy_reward": 0.4072265625, "rewards/brier_reward": 0.7776575922966004, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.40936915278434755, "rewards/smce": -0.023180909082293512, "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": 0.0001953125, "completions/max_length": 851.6, "completions/max_terminated_length": 372.2, "completions/mean_length": 165.8568359375, "completions/mean_terminated_length": 165.5887878417969, "completions/min_length": 76.8, "completions/min_terminated_length": 76.8, "epoch": 0.432, "grad_norm": 0.0009193225996568799, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 453476170.0, "reward": 1.1056634187698364, "reward_std": 0.10354232937097549, "rewards/accuracy_reward": 0.43515625, "rewards/brier_reward": 0.776369285583496, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.41190527081489564, "rewards/smce": -0.042153664771467446, "step": 135 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 496.4, "completions/max_terminated_length": 496.4, "completions/mean_length": 169.11396484375, "completions/mean_terminated_length": 169.11396484375, "completions/min_length": 78.4, "completions/min_terminated_length": 78.4, "epoch": 0.448, "grad_norm": 0.0007092914893291891, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 470160697.0, "reward": 1.0997313261032104, "reward_std": 0.10420847237110138, "rewards/accuracy_reward": 0.4232421875, "rewards/brier_reward": 0.7762229681015015, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4273776412010193, "rewards/smce": -0.03351427987217903, "step": 140 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 472.8, "completions/max_terminated_length": 472.8, "completions/mean_length": 172.30615234375, "completions/mean_terminated_length": 172.30615234375, "completions/min_length": 77.6, "completions/min_terminated_length": 77.6, "epoch": 0.464, "grad_norm": 0.0007953299209475517, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 487095928.0, "reward": 1.0790441513061524, "reward_std": 0.1049978643655777, "rewards/accuracy_reward": 0.37236328125, "rewards/brier_reward": 0.7858292102813721, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4317548990249634, "rewards/smce": -0.07353323474526405, "step": 145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 446.8, "completions/max_terminated_length": 446.8, "completions/mean_length": 172.71796875, "completions/mean_terminated_length": 172.71796875, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.48, "grad_norm": 0.000823770824354142, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 503912592.0, "reward": 1.1081706762313843, "reward_std": 0.10933324098587036, "rewards/accuracy_reward": 0.44482421875, "rewards/brier_reward": 0.7715215682983398, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4173727512359619, "rewards/smce": -0.05264594741165638, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 335.0, "eval_completions/max_terminated_length": 335.0, "eval_completions/mean_length": 171.31009674072266, "eval_completions/mean_terminated_length": 171.31009674072266, "eval_completions/min_length": 89.5, "eval_completions/min_terminated_length": 89.5, "eval_loss": 0.0, "eval_num_tokens": 503912592.0, "eval_reward": 1.0670873522758484, "eval_reward_std": 0.21636013686656952, "eval_rewards/accuracy_reward": 0.341796875, "eval_rewards/brier_reward": 0.7923841774463654, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4122070372104645, "eval_rewards/smce": -0.07125195674598217, "eval_runtime": 12.1739, "eval_samples_per_second": 41.072, "eval_steps_per_second": 0.164, "step": 150 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 892.2, "completions/max_terminated_length": 455.0, "completions/mean_length": 170.50908203125, "completions/mean_terminated_length": 170.24256896972656, "completions/min_length": 82.2, "completions/min_terminated_length": 82.2, "epoch": 0.496, "grad_norm": 0.0009170905104838312, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 520966445.0, "reward": 1.1045333862304687, "reward_std": 0.11436926871538163, "rewards/accuracy_reward": 0.43828125, "rewards/brier_reward": 0.7709831476211548, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4277272403240204, "rewards/smce": -0.031090539321303366, "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": 604.0, "completions/max_terminated_length": 379.4, "completions/mean_length": 169.0560546875, "completions/mean_terminated_length": 168.92233581542968, "completions/min_length": 80.6, "completions/min_terminated_length": 80.6, "epoch": 0.512, "grad_norm": 0.0009231609292328358, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 537843243.0, "reward": 1.1171007394790649, "reward_std": 0.11715430617332459, "rewards/accuracy_reward": 0.45810546875, "rewards/brier_reward": 0.7761961936950683, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44229105710983274, "rewards/smce": -0.033351265173405406, "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": 646.8, "completions/max_terminated_length": 433.0, "completions/mean_length": 170.7927734375, "completions/mean_terminated_length": 170.65943603515626, "completions/min_length": 85.6, "completions/min_terminated_length": 85.6, "epoch": 0.528, "grad_norm": 0.0008310439880006015, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 554621697.0, "reward": 1.1092323064804077, "reward_std": 0.11509672254323959, "rewards/accuracy_reward": 0.43134765625, "rewards/brier_reward": 0.7872170209884644, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44881221652030945, "rewards/smce": -0.032787452824413775, "step": 165 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 846.4, "completions/max_terminated_length": 423.6, "completions/mean_length": 170.37705078125, "completions/mean_terminated_length": 169.9770751953125, "completions/min_length": 81.8, "completions/min_terminated_length": 81.8, "epoch": 0.544, "grad_norm": 0.0007975861080922186, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 571529942.0, "reward": 1.1260975122451782, "reward_std": 0.11977973878383637, "rewards/accuracy_reward": 0.482421875, "rewards/brier_reward": 0.7700685858726501, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4528903365135193, "rewards/smce": -0.03396207746118307, "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": 523.2, "completions/max_terminated_length": 523.2, "completions/mean_length": 167.5654296875, "completions/mean_terminated_length": 167.5654296875, "completions/min_length": 69.8, "completions/min_terminated_length": 69.8, "epoch": 0.56, "grad_norm": 0.0008357908809557557, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 588067220.0, "reward": 1.106752872467041, "reward_std": 0.10880879014730453, "rewards/accuracy_reward": 0.42197265625, "rewards/brier_reward": 0.7916343927383422, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.45748730301856994, "rewards/smce": -0.044184643775224686, "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": 394.4, "completions/max_terminated_length": 394.4, "completions/mean_length": 167.45751953125, "completions/mean_terminated_length": 167.45751953125, "completions/min_length": 78.2, "completions/min_terminated_length": 78.2, "epoch": 0.576, "grad_norm": 0.0007887242827564478, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 604968609.0, "reward": 1.1010312557220459, "reward_std": 0.10108543932437897, "rewards/accuracy_reward": 0.4220703125, "rewards/brier_reward": 0.7800945162773132, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4567448318004608, "rewards/smce": -0.055632703751325605, "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": 0.00068359375, "completions/max_length": 1298.0, "completions/max_terminated_length": 441.0, "completions/mean_length": 165.73251953125, "completions/mean_terminated_length": 164.7956573486328, "completions/min_length": 74.4, "completions/min_terminated_length": 74.4, "epoch": 0.592, "grad_norm": 0.0007809134549461305, "learning_rate": 1e-06, "loss": 0.0016, "num_tokens": 621833422.0, "reward": 1.1107731342315674, "reward_std": 0.10482596158981324, "rewards/accuracy_reward": 0.4439453125, "rewards/brier_reward": 0.7782866358757019, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.46025683879852297, "rewards/smce": -0.0295250304043293, "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": 360.4, "completions/max_terminated_length": 360.4, "completions/mean_length": 162.57333984375, "completions/mean_terminated_length": 162.57333984375, "completions/min_length": 79.2, "completions/min_terminated_length": 79.2, "epoch": 0.608, "grad_norm": 0.000834187027066946, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 638497661.0, "reward": 1.1160224199295044, "reward_std": 0.10795456916093826, "rewards/accuracy_reward": 0.437890625, "rewards/brier_reward": 0.7941561222076416, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.45286591053009034, "rewards/smce": -0.027559639140963554, "step": 190 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 830.6, "completions/max_terminated_length": 486.4, "completions/mean_length": 165.830078125, "completions/mean_terminated_length": 165.56271667480468, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.624, "grad_norm": 0.0009933095425367355, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 655539665.0, "reward": 1.1106883525848388, "reward_std": 0.1104367807507515, "rewards/accuracy_reward": 0.433203125, "rewards/brier_reward": 0.7883719682693482, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4572490334510803, "rewards/smce": -0.039461223781108855, "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": 630.0, "completions/max_terminated_length": 416.0, "completions/mean_length": 169.82666015625, "completions/mean_terminated_length": 169.69326782226562, "completions/min_length": 77.4, "completions/min_terminated_length": 77.4, "epoch": 0.64, "grad_norm": 0.000809962279163301, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 672621378.0, "reward": 1.1414735794067383, "reward_std": 0.10489663779735565, "rewards/accuracy_reward": 0.49716796875, "rewards/brier_reward": 0.785882031917572, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47141573429107664, "rewards/smce": -0.06005169451236725, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 363.5, "eval_completions/max_terminated_length": 363.5, "eval_completions/mean_length": 172.84015655517578, "eval_completions/mean_terminated_length": 172.84015655517578, "eval_completions/min_length": 98.0, "eval_completions/min_terminated_length": 98.0, "eval_loss": 0.0, "eval_num_tokens": 672621378.0, "eval_reward": 1.0921427011489868, "eval_reward_std": 0.2488805055618286, "eval_rewards/accuracy_reward": 0.390625, "eval_rewards/brier_reward": 0.7936670184135437, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4659179747104645, "eval_rewards/smce": -0.07549317367374897, "eval_runtime": 12.161, "eval_samples_per_second": 41.115, "eval_steps_per_second": 0.164, "step": 200 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 434.2, "completions/max_terminated_length": 434.2, "completions/mean_length": 170.987890625, "completions/mean_terminated_length": 170.987890625, "completions/min_length": 82.6, "completions/min_terminated_length": 82.6, "epoch": 0.656, "grad_norm": 0.000862093351315707, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 689228838.0, "reward": 1.0985867977142334, "reward_std": 0.10535258799791336, "rewards/accuracy_reward": 0.4134765625, "rewards/brier_reward": 0.7837995648384094, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.46335841417312623, "rewards/smce": -0.05806147009134292, "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": 604.6, "completions/max_terminated_length": 366.0, "completions/mean_length": 169.04951171875, "completions/mean_terminated_length": 168.91582641601562, "completions/min_length": 81.0, "completions/min_terminated_length": 81.0, "epoch": 0.672, "grad_norm": 0.000558350991923362, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 705873345.0, "reward": 1.1212966680526733, "reward_std": 0.0894002914428711, "rewards/accuracy_reward": 0.45146484375, "rewards/brier_reward": 0.7912296891212464, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.42793651223182677, "rewards/smce": -0.043858605064451696, "step": 210 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 823.8, "completions/max_terminated_length": 454.2, "completions/mean_length": 171.4181640625, "completions/mean_terminated_length": 171.1516357421875, "completions/min_length": 88.0, "completions/min_terminated_length": 88.0, "epoch": 0.688, "grad_norm": 0.0008224369958043098, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 722582587.0, "reward": 1.1220870733261108, "reward_std": 0.09683922380208969, "rewards/accuracy_reward": 0.44853515625, "rewards/brier_reward": 0.7959356069564819, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.43074610233306887, "rewards/smce": -0.04505564570426941, "step": 215 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 647.4, "completions/max_terminated_length": 406.4, "completions/mean_length": 177.95546875, "completions/mean_terminated_length": 177.8226776123047, "completions/min_length": 89.4, "completions/min_terminated_length": 89.4, "epoch": 0.704, "grad_norm": 0.000638193916529417, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 739270995.0, "reward": 1.132350206375122, "reward_std": 0.09680995047092437, "rewards/accuracy_reward": 0.474609375, "rewards/brier_reward": 0.7901916623115539, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44232421517372134, "rewards/smce": -0.03818157017230987, "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": 654.8, "completions/max_terminated_length": 429.0, "completions/mean_length": 181.17880859375, "completions/mean_terminated_length": 181.0465118408203, "completions/min_length": 90.6, "completions/min_terminated_length": 90.6, "epoch": 0.72, "grad_norm": 0.0008984857704490423, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 756136122.0, "reward": 1.1432342529296875, "reward_std": 0.09815936386585236, "rewards/accuracy_reward": 0.48505859375, "rewards/brier_reward": 0.8015105724334717, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.48147950172424314, "rewards/smce": -0.03847266510128975, "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": 691.4, "completions/max_terminated_length": 456.6, "completions/mean_length": 181.17685546875, "completions/mean_terminated_length": 181.04478759765624, "completions/min_length": 90.4, "completions/min_terminated_length": 90.4, "epoch": 0.736, "grad_norm": 0.0009152474231086671, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 772930957.0, "reward": 1.1385881662368775, "reward_std": 0.10552089065313339, "rewards/accuracy_reward": 0.4880859375, "rewards/brier_reward": 0.789192795753479, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5017031252384185, "rewards/smce": -0.057008542120456696, "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": 408.0, "completions/max_terminated_length": 408.0, "completions/mean_length": 182.5458984375, "completions/mean_terminated_length": 182.5458984375, "completions/min_length": 89.4, "completions/min_terminated_length": 89.4, "epoch": 0.752, "grad_norm": 0.0009231127914972603, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 790027427.0, "reward": 1.1306201219558716, "reward_std": 0.10703302025794983, "rewards/accuracy_reward": 0.47080078125, "rewards/brier_reward": 0.79044269323349, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.480672562122345, "rewards/smce": -0.04172721914947033, "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": 702.4, "completions/max_terminated_length": 497.4, "completions/mean_length": 189.3001953125, "completions/mean_terminated_length": 189.16878356933594, "completions/min_length": 95.6, "completions/min_terminated_length": 95.6, "epoch": 0.768, "grad_norm": 0.0008118224795907736, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 806898565.0, "reward": 1.1371404886245728, "reward_std": 0.10227844417095185, "rewards/accuracy_reward": 0.4744140625, "rewards/brier_reward": 0.7999661445617676, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4725029289722443, "rewards/smce": -0.02483233194798231, "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": 443.4, "completions/max_terminated_length": 443.4, "completions/mean_length": 187.94228515625, "completions/mean_terminated_length": 187.94228515625, "completions/min_length": 97.0, "completions/min_terminated_length": 97.0, "epoch": 0.784, "grad_norm": 0.0009792657801881433, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 823997462.0, "reward": 1.1333154678344726, "reward_std": 0.10855860561132431, "rewards/accuracy_reward": 0.47734375, "rewards/brier_reward": 0.7892932295799255, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.49031250476837157, "rewards/smce": -0.06904417928308249, "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": 9.765625e-05, "completions/max_length": 656.8, "completions/max_terminated_length": 427.0, "completions/mean_length": 186.925, "completions/mean_terminated_length": 186.79339599609375, "completions/min_length": 92.0, "completions/min_terminated_length": 92.0, "epoch": 0.8, "grad_norm": 0.0009143622010014951, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 840922134.0, "reward": 1.1496532440185547, "reward_std": 0.1164577692747116, "rewards/accuracy_reward": 0.50009765625, "rewards/brier_reward": 0.7993093371391297, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5264248371124267, "rewards/smce": -0.03849135749042034, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 372.0, "eval_completions/max_terminated_length": 372.0, "eval_completions/mean_length": 188.12112426757812, "eval_completions/mean_terminated_length": 188.12112426757812, "eval_completions/min_length": 123.0, "eval_completions/min_terminated_length": 123.0, "eval_loss": 0.0, "eval_num_tokens": 840922134.0, "eval_reward": 1.0896968841552734, "eval_reward_std": 0.2757375240325928, "eval_rewards/accuracy_reward": 0.390625, "eval_rewards/brier_reward": 0.7887792587280273, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.5048828274011612, "eval_rewards/smce": -0.11445313319563866, "eval_runtime": 12.1809, "eval_samples_per_second": 41.048, "eval_steps_per_second": 0.164, "step": 250 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 644.8, "completions/max_terminated_length": 422.8, "completions/mean_length": 183.82333984375, "completions/mean_terminated_length": 183.69131469726562, "completions/min_length": 95.8, "completions/min_terminated_length": 95.8, "epoch": 0.816, "grad_norm": 0.0011780121130868793, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 857903653.0, "reward": 1.1443308353424073, "reward_std": 0.12021788060665131, "rewards/accuracy_reward": 0.48671875, "rewards/brier_reward": 0.8022388696670533, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.5260625123977661, "rewards/smce": -0.040660728514194486, "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": 651.4, "completions/max_terminated_length": 417.2, "completions/mean_length": 189.85849609375, "completions/mean_terminated_length": 189.72706298828126, "completions/min_length": 88.8, "completions/min_terminated_length": 88.8, "epoch": 0.832, "grad_norm": 0.0010882235364988446, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 874856156.0, "reward": 1.150579810142517, "reward_std": 0.10773497819900513, "rewards/accuracy_reward": 0.49453125, "rewards/brier_reward": 0.8069238066673279, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.49002930521965027, "rewards/smce": -0.03444872796535492, "step": 260 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 889.6, "completions/max_terminated_length": 454.0, "completions/mean_length": 188.8404296875, "completions/mean_terminated_length": 188.57715454101563, "completions/min_length": 101.4, "completions/min_terminated_length": 101.4, "epoch": 0.848, "grad_norm": 0.0009498689323663712, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 891804250.0, "reward": 1.1383338212966918, "reward_std": 0.10923963636159897, "rewards/accuracy_reward": 0.47392578125, "rewards/brier_reward": 0.8029386281967164, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4741074204444885, "rewards/smce": -0.023679725266993044, "step": 265 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 745.0, "completions/max_terminated_length": 551.8, "completions/mean_length": 194.55390625, "completions/mean_terminated_length": 194.42296447753907, "completions/min_length": 102.4, "completions/min_terminated_length": 102.4, "epoch": 0.864, "grad_norm": 0.0009577281307429075, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 908783298.0, "reward": 1.1508663177490235, "reward_std": 0.1009971633553505, "rewards/accuracy_reward": 0.5048828125, "rewards/brier_reward": 0.7971471190452576, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.45706055760383607, "rewards/smce": -0.051518266275525096, "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": 0.0, "completions/max_length": 486.0, "completions/max_terminated_length": 486.0, "completions/mean_length": 197.091015625, "completions/mean_terminated_length": 197.091015625, "completions/min_length": 104.2, "completions/min_terminated_length": 104.2, "epoch": 0.88, "grad_norm": 0.0008310083649121225, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 925948582.0, "reward": 1.1340973615646361, "reward_std": 0.09887916445732117, "rewards/accuracy_reward": 0.45927734375, "rewards/brier_reward": 0.8090185642242431, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.48420898914337157, "rewards/smce": -0.044120116904377935, "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": 658.2, "completions/max_terminated_length": 441.4, "completions/mean_length": 198.308203125, "completions/mean_terminated_length": 198.17782897949218, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.896, "grad_norm": 0.0007895663729868829, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 943090106.0, "reward": 1.138153624534607, "reward_std": 0.10331566035747528, "rewards/accuracy_reward": 0.4728515625, "rewards/brier_reward": 0.8035562992095947, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.49338183999061586, "rewards/smce": -0.038932828418910506, "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": 862.8, "completions/max_terminated_length": 424.4, "completions/mean_length": 196.6095703125, "completions/mean_terminated_length": 196.3483428955078, "completions/min_length": 101.2, "completions/min_terminated_length": 101.2, "epoch": 0.912, "grad_norm": 0.000810799770988524, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 960154684.0, "reward": 1.1308280229568481, "reward_std": 0.09144939780235291, "rewards/accuracy_reward": 0.46455078125, "rewards/brier_reward": 0.7973055481910706, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4340830147266388, "rewards/smce": -0.0579361766576767, "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": 403.4, "completions/max_terminated_length": 403.4, "completions/mean_length": 191.276171875, "completions/mean_terminated_length": 191.276171875, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.928, "grad_norm": 0.0007378010777756572, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 977140168.0, "reward": 1.140564775466919, "reward_std": 0.08266374170780182, "rewards/accuracy_reward": 0.4955078125, "rewards/brier_reward": 0.7856297492980957, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4091914176940918, "rewards/smce": -0.08774990439414979, "step": 290 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 429.0, "completions/max_terminated_length": 429.0, "completions/mean_length": 188.0517578125, "completions/mean_terminated_length": 188.0517578125, "completions/min_length": 98.4, "completions/min_terminated_length": 98.4, "epoch": 0.944, "grad_norm": 0.0008984626037999988, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 994041242.0, "reward": 1.1434009790420532, "reward_std": 0.09434707462787628, "rewards/accuracy_reward": 0.4908203125, "rewards/brier_reward": 0.7959874987602233, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.45385352373123167, "rewards/smce": -0.06694994941353798, "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": 452.2, "completions/max_terminated_length": 452.2, "completions/mean_length": 188.75078125, "completions/mean_terminated_length": 188.75078125, "completions/min_length": 101.8, "completions/min_terminated_length": 101.8, "epoch": 0.96, "grad_norm": 0.0009762116824276745, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 1010914370.0, "reward": 1.1328925371170044, "reward_std": 0.09758208692073822, "rewards/accuracy_reward": 0.46142578125, "rewards/brier_reward": 0.8043647527694702, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5251660048961639, "rewards/smce": -0.06499951593577862, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 424.0, "eval_completions/max_terminated_length": 424.0, "eval_completions/mean_length": 195.71150970458984, "eval_completions/mean_terminated_length": 195.71150970458984, "eval_completions/min_length": 109.5, "eval_completions/min_terminated_length": 109.5, "eval_loss": 0.0, "eval_num_tokens": 1010914370.0, "eval_reward": 1.0907545685768127, "eval_reward_std": 0.29445892572402954, "eval_rewards/accuracy_reward": 0.39453125, "eval_rewards/brier_reward": 0.7869921922683716, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.5476562678813934, "eval_rewards/smce": -0.15322265774011612, "eval_runtime": 13.5384, "eval_samples_per_second": 36.932, "eval_steps_per_second": 0.148, "step": 300 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 411.8, "completions/max_terminated_length": 411.8, "completions/mean_length": 198.3265625, "completions/mean_terminated_length": 198.3265625, "completions/min_length": 103.0, "completions/min_terminated_length": 103.0, "epoch": 0.976, "grad_norm": 0.0008806736441329122, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 1027806354.0, "reward": 1.149838137626648, "reward_std": 0.10467096716165543, "rewards/accuracy_reward": 0.4970703125, "rewards/brier_reward": 0.8026108384132385, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5535937666893005, "rewards/smce": -0.05930444747209549, "step": 305 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 701.0, "completions/max_terminated_length": 556.0, "completions/mean_length": 209.83701171875, "completions/mean_terminated_length": 209.70668029785156, "completions/min_length": 115.0, "completions/min_terminated_length": 115.0, "epoch": 0.992, "grad_norm": 0.0007852715207263827, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 1045083565.0, "reward": 1.1392642974853515, "reward_std": 0.11366586685180664, "rewards/accuracy_reward": 0.483203125, "rewards/brier_reward": 0.7954272270202637, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.500665032863617, "rewards/smce": -0.049921296536922455, "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": 996.0, "completions/max_terminated_length": 477.0, "completions/mean_length": 218.3510284423828, "completions/mean_terminated_length": 218.0293197631836, "completions/min_length": 115.5, "completions/min_terminated_length": 115.5, "epoch": 0.9984, "num_tokens": 1051984536.0, "reward": 1.1309390664100647, "reward_std": 0.1034131683409214, "rewards/accuracy_reward": 0.474609375, "rewards/brier_reward": 0.7875158786773682, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999755859375, "rewards/mean_confidence_reward": 0.509765625, "rewards/smce": -0.03950684051960707, "step": 312, "total_flos": 0.0, "train_loss": 0.004247323407187837, "train_runtime": 42843.6506, "train_samples_per_second": 0.467, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1051984536, "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 }