{ "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": 0.1111328125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1527.8, "completions/mean_length": 468.4916015625, "completions/mean_terminated_length": 335.114697265625, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.004000246990472078, "learning_rate": 3.1249999999999997e-07, "loss": 0.0437, "num_tokens": 19740202.0, "reward": 0.20192502439022064, "reward_std": 0.35805765390396116, "rewards/accuracy_reward": 0.04697265625, "rewards/brier_reward": 0.09627992808818817, "rewards/confidence_one_or_zero": 0.40458984375, "rewards/format_reward": 0.26064453125, "rewards/mean_confidence_reward": 0.536387276649475, "rewards/smce": -0.4894861876964569, "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": 0.09931640625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1529.2, "completions/mean_length": 433.90341796875, "completions/mean_terminated_length": 312.67100830078124, "completions/min_length": 1.2, "completions/min_terminated_length": 1.2, "epoch": 0.032, "grad_norm": 0.004486603196710348, "learning_rate": 6.249999999999999e-07, "loss": 0.05, "num_tokens": 39237421.0, "reward": 0.2321306884288788, "reward_std": 0.38784984946250917, "rewards/accuracy_reward": 0.06201171875, "rewards/brier_reward": 0.1105025827884674, "rewards/confidence_one_or_zero": 0.44853515625, "rewards/format_reward": 0.291796875, "rewards/mean_confidence_reward": 0.5800010561943054, "rewards/smce": -0.5183212876319885, "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": 0.0544921875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1513.8, "completions/mean_length": 315.594140625, "completions/mean_terminated_length": 245.67587585449218, "completions/min_length": 2.4, "completions/min_terminated_length": 2.4, "epoch": 0.048, "grad_norm": 0.005046067293733358, "learning_rate": 9.374999999999999e-07, "loss": 0.0569, "num_tokens": 57570577.0, "reward": 0.4244216620922089, "reward_std": 0.4531022846698761, "rewards/accuracy_reward": 0.12109375, "rewards/brier_reward": 0.2120816856622696, "rewards/confidence_one_or_zero": 0.5412109375, "rewards/format_reward": 0.51572265625, "rewards/mean_confidence_reward": 0.691961121559143, "rewards/smce": -0.572067379951477, "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": 0.0216796875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1420.8, "completions/mean_length": 208.04306640625, "completions/mean_terminated_length": 178.70152587890624, "completions/min_length": 14.6, "completions/min_terminated_length": 14.6, "epoch": 0.064, "grad_norm": 0.002767345402389765, "learning_rate": 1e-06, "loss": 0.0373, "num_tokens": 74636170.0, "reward": 0.6850485920906066, "reward_std": 0.394974547624588, "rewards/accuracy_reward": 0.1884765625, "rewards/brier_reward": 0.33910041451454165, "rewards/confidence_one_or_zero": 0.5734375, "rewards/format_reward": 0.842578125, "rewards/mean_confidence_reward": 0.7928766369819641, "rewards/smce": -0.6067353248596191, "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": 0.01435546875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1331.4, "completions/mean_length": 163.50029296875, "completions/mean_terminated_length": 143.52745666503907, "completions/min_length": 26.2, "completions/min_terminated_length": 26.2, "epoch": 0.08, "grad_norm": 0.00286945141851902, "learning_rate": 1e-06, "loss": 0.026, "num_tokens": 91538797.0, "reward": 0.7999858379364013, "reward_std": 0.32701762318611144, "rewards/accuracy_reward": 0.2197265625, "rewards/brier_reward": 0.42199711203575135, "rewards/confidence_one_or_zero": 0.477734375, "rewards/format_reward": 0.95830078125, "rewards/mean_confidence_reward": 0.7663683414459228, "rewards/smce": -0.5526032984256745, "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": 0.0083984375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1194.4, "completions/mean_length": 137.20341796875, "completions/mean_terminated_length": 125.36412811279297, "completions/min_length": 35.4, "completions/min_terminated_length": 35.4, "epoch": 0.096, "grad_norm": 0.0021565118804574013, "learning_rate": 1e-06, "loss": 0.0144, "num_tokens": 108040208.0, "reward": 0.9100030303001404, "reward_std": 0.2945687472820282, "rewards/accuracy_reward": 0.27177734375, "rewards/brier_reward": 0.5665248811244965, "rewards/confidence_one_or_zero": 0.31298828125, "rewards/format_reward": 0.98173828125, "rewards/mean_confidence_reward": 0.618563997745514, "rewards/smce": -0.36314412355422976, "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": 0.00263671875, "completions/max_length": 1536.0, "completions/max_terminated_length": 791.4, "completions/mean_length": 119.57265625, "completions/mean_terminated_length": 115.82948455810546, "completions/min_length": 45.8, "completions/min_terminated_length": 45.8, "epoch": 0.112, "grad_norm": 0.0013891329290345311, "learning_rate": 1e-06, "loss": 0.0064, "num_tokens": 124112504.0, "reward": 0.9935638666152954, "reward_std": 0.20782245695590973, "rewards/accuracy_reward": 0.2794921875, "rewards/brier_reward": 0.7140938878059387, "rewards/confidence_one_or_zero": 0.18212890625, "rewards/format_reward": 0.9935546875, "rewards/mean_confidence_reward": 0.3873016178607941, "rewards/smce": -0.14182758033275605, "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": 0.00146484375, "completions/max_length": 1536.0, "completions/max_terminated_length": 668.4, "completions/mean_length": 116.70654296875, "completions/mean_terminated_length": 114.6251708984375, "completions/min_length": 28.0, "completions/min_terminated_length": 28.0, "epoch": 0.128, "grad_norm": 0.0009953748667612672, "learning_rate": 1e-06, "loss": 0.0026, "num_tokens": 140292219.0, "reward": 1.029345941543579, "reward_std": 0.14419816434383392, "rewards/accuracy_reward": 0.30634765625, "rewards/brier_reward": 0.7557683944702148, "rewards/confidence_one_or_zero": 0.13935546875, "rewards/format_reward": 0.99658203125, "rewards/mean_confidence_reward": 0.27119520902633665, "rewards/smce": -0.07006132453680039, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00078125, "completions/max_length": 1536.0, "completions/max_terminated_length": 606.0, "completions/mean_length": 115.9509765625, "completions/mean_terminated_length": 114.84084930419922, "completions/min_length": 49.8, "completions/min_terminated_length": 49.8, "epoch": 0.144, "grad_norm": 0.0008894961792975664, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 156452069.0, "reward": 1.0353077411651612, "reward_std": 0.10861439108848572, "rewards/accuracy_reward": 0.28837890625, "rewards/brier_reward": 0.7841946363449097, "rewards/confidence_one_or_zero": 0.07890625, "rewards/format_reward": 0.998046875, "rewards/mean_confidence_reward": 0.24555768072605133, "rewards/smce": -0.05660989247262478, "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": 1308.0, "completions/max_terminated_length": 471.6, "completions/mean_length": 117.7646484375, "completions/mean_terminated_length": 116.51729736328124, "completions/min_length": 50.6, "completions/min_terminated_length": 50.6, "epoch": 0.16, "grad_norm": 0.0008676417637616396, "learning_rate": 1e-06, "loss": 0.0021, "num_tokens": 172609243.0, "reward": 1.0493716478347779, "reward_std": 0.10434378683567047, "rewards/accuracy_reward": 0.33759765625, "rewards/brier_reward": 0.7627172112464905, "rewards/confidence_one_or_zero": 0.04599609375, "rewards/format_reward": 0.9984375, "rewards/mean_confidence_reward": 0.24653425812721252, "rewards/smce": -0.09616437256336212, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.002049180327868827, "eval_completions/max_length": 906.0, "eval_completions/max_terminated_length": 298.0, "eval_completions/mean_length": 121.82882690429688, "eval_completions/mean_terminated_length": 118.9295425415039, "eval_completions/min_length": 62.5, "eval_completions/min_terminated_length": 62.5, "eval_loss": 0.0, "eval_num_tokens": 172609243.0, "eval_reward": 1.026913046836853, "eval_reward_std": 0.15706007927656174, "eval_rewards/accuracy_reward": 0.26171875, "eval_rewards/brier_reward": 0.7960156202316284, "eval_rewards/confidence_one_or_zero": 0.03125, "eval_rewards/format_reward": 0.99609375, "eval_rewards/mean_confidence_reward": 0.2691406309604645, "eval_rewards/smce": -0.026340818963944912, "eval_runtime": 21.7347, "eval_samples_per_second": 23.005, "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": 0.0001953125, "completions/max_length": 854.4, "completions/max_terminated_length": 403.8, "completions/mean_length": 119.148828125, "completions/mean_terminated_length": 118.87189025878907, "completions/min_length": 54.4, "completions/min_terminated_length": 54.4, "epoch": 0.176, "grad_norm": 0.0009692950989119709, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 188849903.0, "reward": 1.0605363368988037, "reward_std": 0.10660427510738373, "rewards/accuracy_reward": 0.36181640625, "rewards/brier_reward": 0.7598503589630127, "rewards/confidence_one_or_zero": 0.02255859375, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.2778467535972595, "rewards/smce": -0.08778789937496186, "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": 0.00068359375, "completions/max_length": 1057.4, "completions/max_terminated_length": 327.0, "completions/mean_length": 122.08525390625, "completions/mean_terminated_length": 121.11732788085938, "completions/min_length": 56.4, "completions/min_terminated_length": 56.4, "epoch": 0.192, "grad_norm": 0.0009788707830011845, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 205073144.0, "reward": 1.0588705778121947, "reward_std": 0.10916559845209121, "rewards/accuracy_reward": 0.35341796875, "rewards/brier_reward": 0.7655003070831299, "rewards/confidence_one_or_zero": 0.0072265625, "rewards/format_reward": 0.998828125, "rewards/mean_confidence_reward": 0.31695576310157775, "rewards/smce": -0.0588741596788168, "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": 944.6, "completions/max_terminated_length": 482.0, "completions/mean_length": 124.083984375, "completions/mean_terminated_length": 123.80798950195313, "completions/min_length": 58.8, "completions/min_terminated_length": 58.8, "epoch": 0.208, "grad_norm": 0.001055966247804463, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 221389876.0, "reward": 1.0805248260498046, "reward_std": 0.12115304917097092, "rewards/accuracy_reward": 0.40693359375, "rewards/brier_reward": 0.7546101808547974, "rewards/confidence_one_or_zero": 0.00390625, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.36039193868637087, "rewards/smce": -0.06622847616672516, "step": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 823.0, "completions/max_terminated_length": 331.6, "completions/mean_length": 126.92802734375, "completions/mean_terminated_length": 126.65286102294922, "completions/min_length": 59.0, "completions/min_terminated_length": 59.0, "epoch": 0.224, "grad_norm": 0.0010693792719393969, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 237640691.0, "reward": 1.0558359622955322, "reward_std": 0.11917430013418198, "rewards/accuracy_reward": 0.33271484375, "rewards/brier_reward": 0.7793541669845581, "rewards/confidence_one_or_zero": 0.0029296875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3942097663879395, "rewards/smce": -0.07278793528676034, "step": 70 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 815.0, "completions/max_terminated_length": 355.2, "completions/mean_length": 130.259765625, "completions/mean_terminated_length": 129.98523559570313, "completions/min_length": 60.8, "completions/min_terminated_length": 60.8, "epoch": 0.24, "grad_norm": 0.0009581490885466337, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 253971927.0, "reward": 1.060979390144348, "reward_std": 0.11544482111930847, "rewards/accuracy_reward": 0.34765625, "rewards/brier_reward": 0.7750885605812072, "rewards/confidence_one_or_zero": 0.0015625, "rewards/format_reward": 0.99921875, "rewards/mean_confidence_reward": 0.3995449125766754, "rewards/smce": -0.05598033778369427, "step": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 591.4, "completions/max_terminated_length": 350.0, "completions/mean_length": 131.16630859375, "completions/mean_terminated_length": 131.029052734375, "completions/min_length": 47.8, "completions/min_terminated_length": 47.8, "epoch": 0.256, "grad_norm": 0.0009065366466529667, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 270476318.0, "reward": 1.0496712446212768, "reward_std": 0.11388081759214401, "rewards/accuracy_reward": 0.3197265625, "rewards/brier_reward": 0.7800132632255554, "rewards/confidence_one_or_zero": 0.00078125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.38173535466194153, "rewards/smce": -0.0739667683839798, "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": 344.2, "completions/max_terminated_length": 344.2, "completions/mean_length": 133.273046875, "completions/mean_terminated_length": 133.273046875, "completions/min_length": 61.0, "completions/min_terminated_length": 61.0, "epoch": 0.272, "grad_norm": 0.000878436432685703, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 286852074.0, "reward": 1.0786157131195069, "reward_std": 0.10490404218435287, "rewards/accuracy_reward": 0.39140625, "rewards/brier_reward": 0.7659255743026734, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.36102074980735777, "rewards/smce": -0.03446613140404224, "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": 640.0, "completions/max_terminated_length": 400.8, "completions/mean_length": 133.86708984375, "completions/mean_terminated_length": 133.73010864257813, "completions/min_length": 60.4, "completions/min_terminated_length": 60.4, "epoch": 0.288, "grad_norm": 0.0008401651284657419, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 303248601.0, "reward": 1.0780799865722657, "reward_std": 0.09450059086084366, "rewards/accuracy_reward": 0.38203125, "rewards/brier_reward": 0.7744266152381897, "rewards/confidence_one_or_zero": 0.00078125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3482990264892578, "rewards/smce": -0.05649419128894806, "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": 575.4, "completions/max_terminated_length": 351.2, "completions/mean_length": 136.93134765625, "completions/mean_terminated_length": 136.79485168457032, "completions/min_length": 63.6, "completions/min_terminated_length": 63.6, "epoch": 0.304, "grad_norm": 0.0008501969859935343, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 319864954.0, "reward": 1.0633502006530762, "reward_std": 0.09683705121278763, "rewards/accuracy_reward": 0.3466796875, "rewards/brier_reward": 0.7802187681198121, "rewards/confidence_one_or_zero": 0.0009765625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3442237675189972, "rewards/smce": -0.03479156941175461, "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": 833.4, "completions/max_terminated_length": 349.0, "completions/mean_length": 139.363671875, "completions/mean_terminated_length": 139.09087829589845, "completions/min_length": 64.4, "completions/min_terminated_length": 64.4, "epoch": 0.32, "grad_norm": 0.0007301330333575606, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 336301126.0, "reward": 1.085144281387329, "reward_std": 0.09936772584915161, "rewards/accuracy_reward": 0.4005859375, "rewards/brier_reward": 0.7700971603393555, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.357449609041214, "rewards/smce": -0.04601215664297342, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 289.0, "eval_completions/max_terminated_length": 289.0, "eval_completions/mean_length": 139.2973175048828, "eval_completions/mean_terminated_length": 139.2973175048828, "eval_completions/min_length": 68.5, "eval_completions/min_terminated_length": 68.5, "eval_loss": 0.0, "eval_num_tokens": 336301126.0, "eval_reward": 1.0518184304237366, "eval_reward_std": 0.18363308906555176, "eval_rewards/accuracy_reward": 0.318359375, "eval_rewards/brier_reward": 0.785280853509903, "eval_rewards/confidence_one_or_zero": 0.001953125, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.3572597801685333, "eval_rewards/smce": -0.04075256921350956, "eval_runtime": 10.943, "eval_samples_per_second": 45.691, "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": 0.0001953125, "completions/max_length": 569.8, "completions/max_terminated_length": 327.0, "completions/mean_length": 139.15380859375, "completions/mean_terminated_length": 138.88138122558593, "completions/min_length": 59.2, "completions/min_terminated_length": 59.2, "epoch": 0.336, "grad_norm": 0.0007222591666504741, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 352403181.0, "reward": 1.075313401222229, "reward_std": 0.1002796545624733, "rewards/accuracy_reward": 0.3818359375, "rewards/brier_reward": 0.7690893411636353, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3662374377250671, "rewards/smce": -0.06171097904443741, "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": 0.000390625, "completions/max_length": 1306.4, "completions/max_terminated_length": 378.6, "completions/mean_length": 139.9060546875, "completions/mean_terminated_length": 139.3603302001953, "completions/min_length": 62.6, "completions/min_terminated_length": 62.6, "epoch": 0.352, "grad_norm": 0.0007559886435046792, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 368906091.0, "reward": 1.089357614517212, "reward_std": 0.09549021869897842, "rewards/accuracy_reward": 0.405859375, "rewards/brier_reward": 0.7732495069503784, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3857435047626495, "rewards/smce": -0.03754127398133278, "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": 619.4, "completions/max_terminated_length": 407.4, "completions/mean_length": 141.80400390625, "completions/mean_terminated_length": 141.66767272949218, "completions/min_length": 63.4, "completions/min_terminated_length": 63.4, "epoch": 0.368, "grad_norm": 0.000884483742993325, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 385270324.0, "reward": 1.0861406326293945, "reward_std": 0.10510718375444413, "rewards/accuracy_reward": 0.39833984375, "rewards/brier_reward": 0.774041223526001, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.39929391741752623, "rewards/smce": -0.028393099457025527, "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": 0.00029296875, "completions/max_length": 1074.6, "completions/max_terminated_length": 350.8, "completions/mean_length": 145.78603515625, "completions/mean_terminated_length": 145.37864685058594, "completions/min_length": 68.6, "completions/min_terminated_length": 68.6, "epoch": 0.384, "grad_norm": 0.0008524227887392044, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 401993605.0, "reward": 1.0776931762695312, "reward_std": 0.11128689795732498, "rewards/accuracy_reward": 0.3865234375, "rewards/brier_reward": 0.7691583395004272, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4166277706623077, "rewards/smce": -0.03233633115887642, "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": 580.6, "completions/max_terminated_length": 349.2, "completions/mean_length": 146.4333984375, "completions/mean_terminated_length": 146.29757995605468, "completions/min_length": 69.0, "completions/min_terminated_length": 69.0, "epoch": 0.4, "grad_norm": 0.0008206322672776878, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 418595739.0, "reward": 1.1109728336334228, "reward_std": 0.10586166381835938, "rewards/accuracy_reward": 0.45146484375, "rewards/brier_reward": 0.7708743572235107, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.43074482679367065, "rewards/smce": -0.037405839934945104, "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": 9.765625e-05, "completions/max_length": 579.6, "completions/max_terminated_length": 356.8, "completions/mean_length": 150.88515625, "completions/mean_terminated_length": 150.7495086669922, "completions/min_length": 66.4, "completions/min_terminated_length": 66.4, "epoch": 0.416, "grad_norm": 0.0009291012538596988, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 435183651.0, "reward": 1.106221318244934, "reward_std": 0.11768909841775894, "rewards/accuracy_reward": 0.44599609375, "rewards/brier_reward": 0.7666454672813415, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4515750050544739, "rewards/smce": -0.04493560083210468, "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": 864.8, "completions/max_terminated_length": 389.6, "completions/mean_length": 156.02626953125, "completions/mean_terminated_length": 155.7564697265625, "completions/min_length": 68.4, "completions/min_terminated_length": 68.4, "epoch": 0.432, "grad_norm": 0.000898000958841294, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 451773584.0, "reward": 1.106620478630066, "reward_std": 0.11107992678880692, "rewards/accuracy_reward": 0.4431640625, "rewards/brier_reward": 0.7702758431434631, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.44765183329582214, "rewards/smce": -0.044996288046240805, "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": 600.0, "completions/max_terminated_length": 399.6, "completions/mean_length": 157.69306640625, "completions/mean_terminated_length": 157.5587921142578, "completions/min_length": 69.0, "completions/min_terminated_length": 69.0, "epoch": 0.448, "grad_norm": 0.0008736915187910199, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 468267273.0, "reward": 1.096074390411377, "reward_std": 0.11458188891410828, "rewards/accuracy_reward": 0.41748046875, "rewards/brier_reward": 0.7748677492141723, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.44305351972579954, "rewards/smce": -0.04980777129530907, "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": 378.2, "completions/max_terminated_length": 378.2, "completions/mean_length": 157.0857421875, "completions/mean_terminated_length": 157.0857421875, "completions/min_length": 68.4, "completions/min_terminated_length": 68.4, "epoch": 0.464, "grad_norm": 0.0008124883752316236, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 484963223.0, "reward": 1.1068456649780274, "reward_std": 0.11299627423286437, "rewards/accuracy_reward": 0.434375, "rewards/brier_reward": 0.7793171048164368, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4311175882816315, "rewards/smce": -0.01590832956135273, "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": 583.2, "completions/max_terminated_length": 345.8, "completions/mean_length": 157.90966796875, "completions/mean_terminated_length": 157.7749816894531, "completions/min_length": 68.6, "completions/min_terminated_length": 68.6, "epoch": 0.48, "grad_norm": 0.0008954492514021695, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 501711034.0, "reward": 1.0894078254699706, "reward_std": 0.10769533514976501, "rewards/accuracy_reward": 0.39990234375, "rewards/brier_reward": 0.7791118502616883, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.43644275069236754, "rewards/smce": -0.040997632779181004, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 290.0, "eval_completions/max_terminated_length": 290.0, "eval_completions/mean_length": 157.6277847290039, "eval_completions/mean_terminated_length": 157.6277847290039, "eval_completions/min_length": 85.0, "eval_completions/min_terminated_length": 85.0, "eval_loss": 0.0, "eval_num_tokens": 501711034.0, "eval_reward": 1.0702556371688843, "eval_reward_std": 0.2216203734278679, "eval_rewards/accuracy_reward": 0.35546875, "eval_rewards/brier_reward": 0.7850488126277924, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4248046875, "eval_rewards/smce": -0.07097656466066837, "eval_runtime": 10.9948, "eval_samples_per_second": 45.476, "eval_steps_per_second": 0.182, "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": 0.00029296875, "completions/max_length": 1074.8, "completions/max_terminated_length": 377.8, "completions/mean_length": 158.95439453125, "completions/mean_terminated_length": 158.55074157714844, "completions/min_length": 69.0, "completions/min_terminated_length": 69.0, "epoch": 0.496, "grad_norm": 0.000837022322230041, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 518148647.0, "reward": 1.1130841255187989, "reward_std": 0.10837481170892715, "rewards/accuracy_reward": 0.45146484375, "rewards/brier_reward": 0.7749992847442627, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.439496785402298, "rewards/smce": -0.03746632831171155, "step": 155 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 371.0, "completions/max_terminated_length": 371.0, "completions/mean_length": 161.6998046875, "completions/mean_terminated_length": 161.6998046875, "completions/min_length": 67.6, "completions/min_terminated_length": 67.6, "epoch": 0.512, "grad_norm": 0.0007171882316470146, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 534711397.0, "reward": 1.0941582918167114, "reward_std": 0.09756620526313782, "rewards/accuracy_reward": 0.40478515625, "rewards/brier_reward": 0.7835341572761536, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4164893627166748, "rewards/smce": -0.03459956496953964, "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": 615.2, "completions/max_terminated_length": 392.8, "completions/mean_length": 165.54208984375, "completions/mean_terminated_length": 165.4082794189453, "completions/min_length": 75.2, "completions/min_terminated_length": 75.2, "epoch": 0.528, "grad_norm": 0.0009193150326609612, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 551478228.0, "reward": 1.1048997163772583, "reward_std": 0.10788022428750992, "rewards/accuracy_reward": 0.42314453125, "rewards/brier_reward": 0.7867572665214538, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4206361770629883, "rewards/smce": -0.05594331808388233, "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": 1069.4, "completions/max_terminated_length": 372.4, "completions/mean_length": 164.344140625, "completions/mean_terminated_length": 163.94191284179686, "completions/min_length": 79.0, "completions/min_terminated_length": 79.0, "epoch": 0.544, "grad_norm": 0.0009100640309043229, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 568289304.0, "reward": 1.1251786947250366, "reward_std": 0.1073757141828537, "rewards/accuracy_reward": 0.4765625, "rewards/brier_reward": 0.7740918874740601, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.44392098784446715, "rewards/smce": -0.04849136099219322, "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": 0.0001953125, "completions/max_length": 864.6, "completions/max_terminated_length": 431.4, "completions/mean_length": 167.23515625, "completions/mean_terminated_length": 166.96788024902344, "completions/min_length": 79.6, "completions/min_terminated_length": 79.6, "epoch": 0.56, "grad_norm": 0.0008807600825093687, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 585010496.0, "reward": 1.1237701177597046, "reward_std": 0.10904805213212967, "rewards/accuracy_reward": 0.465234375, "rewards/brier_reward": 0.7825065374374389, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4573706090450287, "rewards/smce": -0.06266507729887963, "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": 0.0001953125, "completions/max_length": 841.8, "completions/max_terminated_length": 410.6, "completions/mean_length": 171.80546875, "completions/mean_terminated_length": 171.5396240234375, "completions/min_length": 84.2, "completions/min_terminated_length": 84.2, "epoch": 0.576, "grad_norm": 0.000868572446051985, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 601983000.0, "reward": 1.120293092727661, "reward_std": 0.10446289032697678, "rewards/accuracy_reward": 0.44287109375, "rewards/brier_reward": 0.7979134917259216, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4504714846611023, "rewards/smce": -0.03961756564676762, "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": 9.765625e-05, "completions/max_length": 719.2, "completions/max_terminated_length": 499.2, "completions/mean_length": 178.27021484375, "completions/mean_terminated_length": 178.13722534179686, "completions/min_length": 89.2, "completions/min_terminated_length": 89.2, "epoch": 0.592, "grad_norm": 0.0007287701591849327, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 618830951.0, "reward": 1.0865333557128907, "reward_std": 0.1005751445889473, "rewards/accuracy_reward": 0.38828125, "rewards/brier_reward": 0.7848896384239197, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4494062900543213, "rewards/smce": -0.07449966222047806, "step": 185 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 643.8, "completions/max_terminated_length": 412.2, "completions/mean_length": 180.780859375, "completions/mean_terminated_length": 180.64830932617187, "completions/min_length": 86.8, "completions/min_terminated_length": 86.8, "epoch": 0.608, "grad_norm": 0.000843865389470011, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 635613475.0, "reward": 1.1116358280181884, "reward_std": 0.0987422376871109, "rewards/accuracy_reward": 0.43408203125, "rewards/brier_reward": 0.7892889738082886, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.43660807609558105, "rewards/smce": -0.025845246762037276, "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": 946.4, "completions/max_terminated_length": 506.8, "completions/mean_length": 180.91884765625, "completions/mean_terminated_length": 180.65436096191405, "completions/min_length": 88.4, "completions/min_terminated_length": 88.4, "epoch": 0.624, "grad_norm": 0.0007394336280412972, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 652432676.0, "reward": 1.1127588748931885, "reward_std": 0.09222938418388367, "rewards/accuracy_reward": 0.4365234375, "rewards/brier_reward": 0.7891914367675781, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.41735615730285647, "rewards/smce": -0.02638731971383095, "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": 706.4, "completions/max_terminated_length": 476.2, "completions/mean_length": 179.44296875, "completions/mean_terminated_length": 179.31036071777345, "completions/min_length": 83.0, "completions/min_terminated_length": 83.0, "epoch": 0.64, "grad_norm": 0.0009229676797986031, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 669289564.0, "reward": 1.1244325637817383, "reward_std": 0.10907497704029083, "rewards/accuracy_reward": 0.463671875, "rewards/brier_reward": 0.7852943778038025, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4432168066501617, "rewards/smce": -0.04280035831034183, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 344.5, "eval_completions/max_terminated_length": 344.5, "eval_completions/mean_length": 178.71320343017578, "eval_completions/mean_terminated_length": 178.71320343017578, "eval_completions/min_length": 103.0, "eval_completions/min_terminated_length": 103.0, "eval_loss": 0.0, "eval_num_tokens": 669289564.0, "eval_reward": 1.074850082397461, "eval_reward_std": 0.24065522849559784, "eval_rewards/accuracy_reward": 0.359375, "eval_rewards/brier_reward": 0.7903341948986053, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.45826172828674316, "eval_rewards/smce": -0.09919121861457825, "eval_runtime": 11.9627, "eval_samples_per_second": 41.797, "eval_steps_per_second": 0.167, "step": 200 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 665.4, "completions/max_terminated_length": 434.0, "completions/mean_length": 175.375, "completions/mean_terminated_length": 175.24189147949218, "completions/min_length": 82.6, "completions/min_terminated_length": 82.6, "epoch": 0.656, "grad_norm": 0.0008042529225349426, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 686066204.0, "reward": 1.11544828414917, "reward_std": 0.11178317070007324, "rewards/accuracy_reward": 0.448046875, "rewards/brier_reward": 0.7829502701759339, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4799645602703094, "rewards/smce": -0.03865515626966953, "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": 648.8, "completions/max_terminated_length": 424.0, "completions/mean_length": 172.976953125, "completions/mean_terminated_length": 172.84388122558593, "completions/min_length": 82.8, "completions/min_terminated_length": 82.8, "epoch": 0.672, "grad_norm": 0.000913208001293242, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 702852176.0, "reward": 1.1459183931350707, "reward_std": 0.11151646673679352, "rewards/accuracy_reward": 0.50302734375, "rewards/brier_reward": 0.7889090418815613, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4935117304325104, "rewards/smce": -0.02932633813470602, "step": 210 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 533.8, "completions/max_terminated_length": 533.8, "completions/mean_length": 169.51435546875, "completions/mean_terminated_length": 169.51435546875, "completions/min_length": 83.4, "completions/min_terminated_length": 83.4, "epoch": 0.688, "grad_norm": 0.0009185533272102475, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 719669411.0, "reward": 1.11937415599823, "reward_std": 0.1056325763463974, "rewards/accuracy_reward": 0.445703125, "rewards/brier_reward": 0.7930505633354187, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4931591808795929, "rewards/smce": -0.0630765751004219, "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": 0.0001953125, "completions/max_length": 830.6, "completions/max_terminated_length": 371.4, "completions/mean_length": 170.2216796875, "completions/mean_terminated_length": 169.9547332763672, "completions/min_length": 84.6, "completions/min_terminated_length": 84.6, "epoch": 0.704, "grad_norm": 0.0010203915880993009, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 736362849.0, "reward": 1.1581217765808105, "reward_std": 0.11780242323875427, "rewards/accuracy_reward": 0.53212890625, "rewards/brier_reward": 0.78431156873703, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5082412004470825, "rewards/smce": -0.026287396252155305, "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": 669.2, "completions/max_terminated_length": 448.0, "completions/mean_length": 173.9283203125, "completions/mean_terminated_length": 173.79610290527344, "completions/min_length": 87.6, "completions/min_terminated_length": 87.6, "epoch": 0.72, "grad_norm": 0.0008801869698800147, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 753279939.0, "reward": 1.1250783205032349, "reward_std": 0.11430736482143403, "rewards/accuracy_reward": 0.4568359375, "rewards/brier_reward": 0.7934238076210022, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.49130371809005735, "rewards/smce": -0.06444230675697327, "step": 225 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 849.2, "completions/max_terminated_length": 405.2, "completions/mean_length": 174.85498046875, "completions/mean_terminated_length": 174.58867492675782, "completions/min_length": 82.8, "completions/min_terminated_length": 82.8, "epoch": 0.736, "grad_norm": 0.0009612944559194148, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 770042742.0, "reward": 1.1493725538253785, "reward_std": 0.10344905704259873, "rewards/accuracy_reward": 0.5033203125, "rewards/brier_reward": 0.7956208944320678, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5045771479606629, "rewards/smce": -0.018064015358686448, "step": 230 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 644.2, "completions/max_terminated_length": 427.4, "completions/mean_length": 179.42392578125, "completions/mean_terminated_length": 179.29153442382812, "completions/min_length": 87.8, "completions/min_terminated_length": 87.8, "epoch": 0.752, "grad_norm": 0.0010464686201885343, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 786916075.0, "reward": 1.1549260854721068, "reward_std": 0.11921477615833283, "rewards/accuracy_reward": 0.5169921875, "rewards/brier_reward": 0.7929610848426819, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5239619135856628, "rewards/smce": -0.04414962157607079, "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": 799.4, "completions/max_terminated_length": 617.8, "completions/mean_length": 184.51640625, "completions/mean_terminated_length": 184.38465270996093, "completions/min_length": 87.4, "completions/min_terminated_length": 87.4, "epoch": 0.768, "grad_norm": 0.0009205514215864241, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 804057267.0, "reward": 1.1374267578125, "reward_std": 0.10838203877210617, "rewards/accuracy_reward": 0.4748046875, "rewards/brier_reward": 0.800150191783905, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5018200159072876, "rewards/smce": -0.04639698397368193, "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": 458.6, "completions/max_terminated_length": 458.6, "completions/mean_length": 183.49248046875, "completions/mean_terminated_length": 183.49248046875, "completions/min_length": 85.4, "completions/min_terminated_length": 85.4, "epoch": 0.784, "grad_norm": 0.0008747904212214053, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 820903718.0, "reward": 1.140303111076355, "reward_std": 0.1141647920012474, "rewards/accuracy_reward": 0.493359375, "rewards/brier_reward": 0.7873458147048951, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.49045605063438413, "rewards/smce": -0.02265973724424839, "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": 0.0001953125, "completions/max_length": 917.6, "completions/max_terminated_length": 471.4, "completions/mean_length": 185.45810546875, "completions/mean_terminated_length": 185.19440307617188, "completions/min_length": 96.2, "completions/min_terminated_length": 96.2, "epoch": 0.8, "grad_norm": 0.0008962493157014251, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 837924505.0, "reward": 1.1438107013702392, "reward_std": 0.10105291306972504, "rewards/accuracy_reward": 0.488671875, "rewards/brier_reward": 0.7992471814155578, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.47882911562919617, "rewards/smce": -0.056229662522673605, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 354.5, "eval_completions/max_terminated_length": 354.5, "eval_completions/mean_length": 188.99618530273438, "eval_completions/mean_terminated_length": 188.99618530273438, "eval_completions/min_length": 108.5, "eval_completions/min_terminated_length": 108.5, "eval_loss": 0.0, "eval_num_tokens": 837924505.0, "eval_reward": 1.0744569301605225, "eval_reward_std": 0.2566024512052536, "eval_rewards/accuracy_reward": 0.345703125, "eval_rewards/brier_reward": 0.80322265625, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.47246094048023224, "eval_rewards/smce": -0.12773926183581352, "eval_runtime": 13.1459, "eval_samples_per_second": 38.035, "eval_steps_per_second": 0.152, "step": 250 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 463.6, "completions/max_terminated_length": 463.6, "completions/mean_length": 187.904296875, "completions/mean_terminated_length": 187.904296875, "completions/min_length": 94.8, "completions/min_terminated_length": 94.8, "epoch": 0.816, "grad_norm": 0.0009017204283736646, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 855033413.0, "reward": 1.1587502717971803, "reward_std": 0.09949774891138077, "rewards/accuracy_reward": 0.50849609375, "rewards/brier_reward": 0.8090077996253967, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4960273504257202, "rewards/smce": -0.04275580998510122, "step": 255 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 560.0, "completions/max_terminated_length": 560.0, "completions/mean_length": 194.63994140625, "completions/mean_terminated_length": 194.63994140625, "completions/min_length": 100.4, "completions/min_terminated_length": 100.4, "epoch": 0.832, "grad_norm": 0.000874867953825742, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 872085822.0, "reward": 1.1352928638458253, "reward_std": 0.10674976855516434, "rewards/accuracy_reward": 0.47353515625, "rewards/brier_reward": 0.7970545411109924, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.505438482761383, "rewards/smce": -0.04949844591319561, "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": 941.2, "completions/max_terminated_length": 491.8, "completions/mean_length": 198.4744140625, "completions/mean_terminated_length": 198.21253662109376, "completions/min_length": 105.2, "completions/min_terminated_length": 105.2, "epoch": 0.848, "grad_norm": 0.0008845736738294363, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 889214136.0, "reward": 1.1375447273254395, "reward_std": 0.09591394364833832, "rewards/accuracy_reward": 0.47041015625, "rewards/brier_reward": 0.8048787593841553, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5062236309051513, "rewards/smce": -0.0516842994838953, "step": 265 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 509.0, "completions/max_terminated_length": 509.0, "completions/mean_length": 199.49775390625, "completions/mean_terminated_length": 199.49775390625, "completions/min_length": 97.4, "completions/min_terminated_length": 97.4, "epoch": 0.864, "grad_norm": 0.0005732330610044301, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 906454049.0, "reward": 1.1363605737686158, "reward_std": 0.09182919263839721, "rewards/accuracy_reward": 0.4693359375, "rewards/brier_reward": 0.8035852432250976, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4373330116271973, "rewards/smce": -0.05709770061075688, "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": 683.2, "completions/max_terminated_length": 484.0, "completions/mean_length": 206.20185546875, "completions/mean_terminated_length": 206.07252502441406, "completions/min_length": 105.6, "completions/min_terminated_length": 105.6, "epoch": 0.88, "grad_norm": 0.0007916499162092805, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 923661428.0, "reward": 1.128675103187561, "reward_std": 0.09593857079744339, "rewards/accuracy_reward": 0.47060546875, "rewards/brier_reward": 0.7869455933570861, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4107128918170929, "rewards/smce": -0.06385259106755256, "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": 714.8, "completions/max_terminated_length": 508.6, "completions/mean_length": 206.65712890625, "completions/mean_terminated_length": 206.5268341064453, "completions/min_length": 105.6, "completions/min_terminated_length": 105.6, "epoch": 0.896, "grad_norm": 0.0008560984279029071, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 940891261.0, "reward": 1.1234555006027223, "reward_std": 0.10874682813882827, "rewards/accuracy_reward": 0.45595703125, "rewards/brier_reward": 0.7910539627075195, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4712998032569885, "rewards/smce": -0.033051177859306335, "step": 280 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 761.2, "completions/max_terminated_length": 551.4, "completions/mean_length": 200.74912109375, "completions/mean_terminated_length": 200.61900634765624, "completions/min_length": 100.4, "completions/min_terminated_length": 100.4, "epoch": 0.912, "grad_norm": 0.0010159144876524806, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 957992660.0, "reward": 1.1473841905593871, "reward_std": 0.10081676095724106, "rewards/accuracy_reward": 0.49599609375, "rewards/brier_reward": 0.7989710807800293, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4727851629257202, "rewards/smce": -0.04406613744795322, "step": 285 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 691.6, "completions/max_terminated_length": 482.4, "completions/mean_length": 191.688671875, "completions/mean_terminated_length": 191.4256622314453, "completions/min_length": 96.4, "completions/min_terminated_length": 96.4, "epoch": 0.928, "grad_norm": 0.0008976598037406802, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 974623840.0, "reward": 1.1475050926208497, "reward_std": 0.0983053594827652, "rewards/accuracy_reward": 0.50693359375, "rewards/brier_reward": 0.7882754921913147, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.48968067169189455, "rewards/smce": -0.04574570134282112, "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": 695.0, "completions/max_terminated_length": 478.4, "completions/mean_length": 191.94453125, "completions/mean_terminated_length": 191.81333618164064, "completions/min_length": 98.2, "completions/min_terminated_length": 98.2, "epoch": 0.944, "grad_norm": 0.0008249140810221434, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 991602376.0, "reward": 1.1368897676467895, "reward_std": 0.09058477282524109, "rewards/accuracy_reward": 0.4677734375, "rewards/brier_reward": 0.8061062097549438, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4747841954231262, "rewards/smce": -0.03453637734055519, "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": 680.2, "completions/max_terminated_length": 487.2, "completions/mean_length": 192.143359375, "completions/mean_terminated_length": 192.01217956542968, "completions/min_length": 92.4, "completions/min_terminated_length": 92.4, "epoch": 0.96, "grad_norm": 0.0008051492040976882, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 1008719364.0, "reward": 1.1513455629348754, "reward_std": 0.0891915887594223, "rewards/accuracy_reward": 0.49482421875, "rewards/brier_reward": 0.8079674720764161, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4651210904121399, "rewards/smce": -0.03700222373008728, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 372.0, "eval_completions/max_terminated_length": 372.0, "eval_completions/mean_length": 192.57955932617188, "eval_completions/mean_terminated_length": 192.57955932617188, "eval_completions/min_length": 103.5, "eval_completions/min_terminated_length": 103.5, "eval_loss": 0.0, "eval_num_tokens": 1008719364.0, "eval_reward": 1.0884253978729248, "eval_reward_std": 0.2481953427195549, "eval_rewards/accuracy_reward": 0.369140625, "eval_rewards/brier_reward": 0.8077171742916107, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.44535157084465027, "eval_rewards/smce": -0.07722852565348148, "eval_runtime": 12.6348, "eval_samples_per_second": 39.573, "eval_steps_per_second": 0.158, "step": 300 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1102.6, "completions/max_terminated_length": 473.4, "completions/mean_length": 192.79931640625, "completions/mean_terminated_length": 192.40595397949218, "completions/min_length": 102.4, "completions/min_terminated_length": 102.4, "epoch": 0.976, "grad_norm": 0.0008862409740686417, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 1025823645.0, "reward": 1.1745065450668335, "reward_std": 0.08961422890424728, "rewards/accuracy_reward": 0.53203125, "rewards/brier_reward": 0.817278528213501, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.48915820717811587, "rewards/smce": -0.04640830308198929, "step": 305 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 483.8, "completions/max_terminated_length": 483.8, "completions/mean_length": 196.373828125, "completions/mean_terminated_length": 196.373828125, "completions/min_length": 104.2, "completions/min_terminated_length": 104.2, "epoch": 0.992, "grad_norm": 0.0009224271052516997, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 1042824881.0, "reward": 1.1565107822418212, "reward_std": 0.08400667905807495, "rewards/accuracy_reward": 0.50283203125, "rewards/brier_reward": 0.8101913809776307, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5155996084213257, "rewards/smce": -0.02968781553208828, "step": 310 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 1057.5, "completions/max_terminated_length": 501.5, "completions/mean_length": 202.44482421875, "completions/mean_terminated_length": 201.7928466796875, "completions/min_length": 97.0, "completions/min_terminated_length": 97.0, "epoch": 0.9984, "num_tokens": 1049742511.0, "reward": 1.1453857421875, "reward_std": 0.09386108070611954, "rewards/accuracy_reward": 0.48779296875, "rewards/brier_reward": 0.8034725785255432, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.49439698457717896, "rewards/smce": -0.06732712313532829, "step": 312, "total_flos": 0.0, "train_loss": 0.00036888672133127094, "train_runtime": 33617.5864, "train_samples_per_second": 0.595, "train_steps_per_second": 0.009 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1049742511, "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 }