{ "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": 0.1095703125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1532.6, "completions/mean_length": 468.423046875, "completions/mean_terminated_length": 337.1365539550781, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.0029375546146184206, "learning_rate": 3.1249999999999997e-07, "loss": 0.0267, "num_tokens": 19640684.0, "reward": -0.016939924284815788, "reward_std": 0.4662073254585266, "rewards/accuracy_reward": 0.0642578125, "rewards/brier_reward": 0.11008564978837967, "rewards/confidence_one_or_zero": 0.41435546875, "rewards/format_reward": 0.2634765625, "rewards/mean_confidence_reward": 0.5356804013252259, "rewards/smce": -0.4717017710208893, "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": 0.10654296875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1530.8, "completions/mean_length": 456.72236328125, "completions/mean_terminated_length": 328.05751953125, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.003135639475658536, "learning_rate": 6.249999999999999e-07, "loss": 0.0229, "num_tokens": 39417873.0, "reward": -0.0207392705604434, "reward_std": 0.4668224513530731, "rewards/accuracy_reward": 0.05888671875, "rewards/brier_reward": 0.11468577980995179, "rewards/confidence_one_or_zero": 0.428125, "rewards/format_reward": 0.27978515625, "rewards/mean_confidence_reward": 0.5534478187561035, "rewards/smce": -0.49483814239501955, "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": 0.06904296875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1517.0, "completions/mean_length": 344.80185546875, "completions/mean_terminated_length": 257.0526123046875, "completions/min_length": 1.2, "completions/min_terminated_length": 1.2, "epoch": 0.048, "grad_norm": 0.0045189037919044495, "learning_rate": 9.374999999999999e-07, "loss": 0.0296, "num_tokens": 57997380.0, "reward": 0.10179619886912405, "reward_std": 0.5125089168548584, "rewards/accuracy_reward": 0.0970703125, "rewards/brier_reward": 0.18347085118293763, "rewards/confidence_one_or_zero": 0.48662109375, "rewards/format_reward": 0.45732421875, "rewards/mean_confidence_reward": 0.6305847048759461, "rewards/smce": -0.534275221824646, "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": 0.0271484375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1432.0, "completions/mean_length": 220.9154296875, "completions/mean_terminated_length": 184.3703186035156, "completions/min_length": 2.6, "completions/min_terminated_length": 2.6, "epoch": 0.064, "grad_norm": 0.00436444440856576, "learning_rate": 1e-06, "loss": 0.027, "num_tokens": 75177954.0, "reward": 0.40185180902481077, "reward_std": 0.5494400143623352, "rewards/accuracy_reward": 0.18125, "rewards/brier_reward": 0.36000813245773317, "rewards/confidence_one_or_zero": 0.5275390625, "rewards/format_reward": 0.81787109375, "rewards/mean_confidence_reward": 0.732852327823639, "rewards/smce": -0.5554281115531922, "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": 0.0171875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1267.0, "completions/mean_length": 173.53408203125, "completions/mean_terminated_length": 149.69837951660156, "completions/min_length": 7.4, "completions/min_terminated_length": 7.4, "epoch": 0.08, "grad_norm": 0.004786408040672541, "learning_rate": 1e-06, "loss": 0.0181, "num_tokens": 91888095.0, "reward": 0.6115968227386475, "reward_std": 0.5032546579837799, "rewards/accuracy_reward": 0.209765625, "rewards/brier_reward": 0.5036922097206116, "rewards/confidence_one_or_zero": 0.43955078125, "rewards/format_reward": 0.9505859375, "rewards/mean_confidence_reward": 0.6410665035247802, "rewards/smce": -0.4408522069454193, "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": 0.00849609375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1187.0, "completions/mean_length": 147.68828125, "completions/mean_terminated_length": 135.78773193359376, "completions/min_length": 5.8, "completions/min_terminated_length": 5.8, "epoch": 0.096, "grad_norm": 0.001847111969254911, "learning_rate": 1e-06, "loss": 0.0116, "num_tokens": 108445031.0, "reward": 0.8734211802482605, "reward_std": 0.3274820476770401, "rewards/accuracy_reward": 0.2228515625, "rewards/brier_reward": 0.7075286149978638, "rewards/confidence_one_or_zero": 0.415234375, "rewards/format_reward": 0.9798828125, "rewards/mean_confidence_reward": 0.31975880861282346, "rewards/smce": -0.1634220853447914, "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": 0.00546875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1032.6, "completions/mean_length": 129.9931640625, "completions/mean_terminated_length": 122.26376342773438, "completions/min_length": 33.2, "completions/min_terminated_length": 33.2, "epoch": 0.112, "grad_norm": 0.0019039362668991089, "learning_rate": 1e-06, "loss": 0.0101, "num_tokens": 124885633.0, "reward": 0.9506356954574585, "reward_std": 0.16324757039546967, "rewards/accuracy_reward": 0.23447265625, "rewards/brier_reward": 0.7872580051422119, "rewards/confidence_one_or_zero": 0.47919921875, "rewards/format_reward": 0.99111328125, "rewards/mean_confidence_reward": 0.1406856507062912, "rewards/smce": -0.11157373487949371, "step": 35 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0017578125, "completions/max_length": 1536.0, "completions/max_terminated_length": 831.4, "completions/mean_length": 122.6791015625, "completions/mean_terminated_length": 120.19132690429687, "completions/min_length": 33.2, "completions/min_terminated_length": 33.2, "epoch": 0.128, "grad_norm": 0.001967265736311674, "learning_rate": 1e-06, "loss": 0.003, "num_tokens": 141058539.0, "reward": 0.9794683218002319, "reward_std": 0.14601433873176575, "rewards/accuracy_reward": 0.20146484375, "rewards/brier_reward": 0.8251733541488647, "rewards/confidence_one_or_zero": 0.3142578125, "rewards/format_reward": 0.99560546875, "rewards/mean_confidence_reward": 0.16780101656913757, "rewards/smce": -0.06330799236893654, "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": 0.0015625, "completions/max_length": 1332.0, "completions/max_terminated_length": 777.4, "completions/mean_length": 121.993359375, "completions/mean_terminated_length": 119.779833984375, "completions/min_length": 39.8, "completions/min_terminated_length": 39.8, "epoch": 0.144, "grad_norm": 0.002123710699379444, "learning_rate": 1e-06, "loss": 0.0021, "num_tokens": 157258183.0, "reward": 1.0028710842132569, "reward_std": 0.19501619338989257, "rewards/accuracy_reward": 0.20185546875, "rewards/brier_reward": 0.8341439127922058, "rewards/confidence_one_or_zero": 0.13662109375, "rewards/format_reward": 0.99736328125, "rewards/mean_confidence_reward": 0.21607398390769958, "rewards/smce": -0.027621227875351907, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00107421875, "completions/max_length": 1536.0, "completions/max_terminated_length": 457.8, "completions/mean_length": 123.31552734375, "completions/mean_terminated_length": 121.7956298828125, "completions/min_length": 44.4, "completions/min_terminated_length": 44.4, "epoch": 0.16, "grad_norm": 0.0010736105032265186, "learning_rate": 1e-06, "loss": 0.002, "num_tokens": 173541862.0, "reward": 1.0139008045196534, "reward_std": 0.13272683173418046, "rewards/accuracy_reward": 0.22783203125, "rewards/brier_reward": 0.8250125885009766, "rewards/confidence_one_or_zero": 0.058203125, "rewards/format_reward": 0.9982421875, "rewards/mean_confidence_reward": 0.22021350860595704, "rewards/smce": -0.023285704664885998, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 387.0, "eval_completions/max_terminated_length": 387.0, "eval_completions/mean_length": 122.21080780029297, "eval_completions/mean_terminated_length": 122.21080780029297, "eval_completions/min_length": 64.0, "eval_completions/min_terminated_length": 64.0, "eval_loss": 0.0, "eval_num_tokens": 173541862.0, "eval_reward": 0.9996126294136047, "eval_reward_std": 0.19382363185286522, "eval_rewards/accuracy_reward": 0.19921875, "eval_rewards/brier_reward": 0.8393352031707764, "eval_rewards/confidence_one_or_zero": 0.04296875, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.2296484410762787, "eval_rewards/smce": -0.03932909946888685, "eval_runtime": 12.2354, "eval_samples_per_second": 40.865, "eval_steps_per_second": 0.163, "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": 0.00048828125, "completions/max_length": 1251.8, "completions/max_terminated_length": 583.6, "completions/mean_length": 121.42333984375, "completions/mean_terminated_length": 120.73204345703125, "completions/min_length": 44.6, "completions/min_terminated_length": 44.6, "epoch": 0.176, "grad_norm": 0.001155526377260685, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 190022357.0, "reward": 1.0153558492660522, "reward_std": 0.12880419939756393, "rewards/accuracy_reward": 0.24404296875, "rewards/brier_reward": 0.8160562515258789, "rewards/confidence_one_or_zero": 0.02265625, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.2302953004837036, "rewards/smce": -0.02841135747730732, "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": 1041.8, "completions/max_terminated_length": 362.0, "completions/mean_length": 119.63798828125, "completions/mean_terminated_length": 118.9463104248047, "completions/min_length": 45.6, "completions/min_terminated_length": 45.6, "epoch": 0.192, "grad_norm": 0.0021765059791505337, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 206062266.0, "reward": 1.0223777770996094, "reward_std": 0.1203025907278061, "rewards/accuracy_reward": 0.28935546875, "rewards/brier_reward": 0.7992676138877869, "rewards/confidence_one_or_zero": 0.00576171875, "rewards/format_reward": 0.99873046875, "rewards/mean_confidence_reward": 0.27211748361587523, "rewards/smce": -0.042598482221364975, "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": 916.4, "completions/max_terminated_length": 529.2, "completions/mean_length": 120.6912109375, "completions/mean_terminated_length": 120.41455688476563, "completions/min_length": 57.0, "completions/min_terminated_length": 57.0, "epoch": 0.208, "grad_norm": 0.0020183834712952375, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 222330368.0, "reward": 1.0374894618988038, "reward_std": 0.17807672172784805, "rewards/accuracy_reward": 0.31611328125, "rewards/brier_reward": 0.788511848449707, "rewards/confidence_one_or_zero": 0.0029296875, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.3276663362979889, "rewards/smce": -0.02906099855899811, "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": 9.765625e-05, "completions/max_length": 801.8, "completions/max_terminated_length": 576.8, "completions/mean_length": 118.653515625, "completions/mean_terminated_length": 118.5151382446289, "completions/min_length": 57.0, "completions/min_terminated_length": 57.0, "epoch": 0.224, "grad_norm": 0.0018503839382901788, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 238698564.0, "reward": 1.0381819605827332, "reward_std": 0.19111664593219757, "rewards/accuracy_reward": 0.318359375, "rewards/brier_reward": 0.7878335952758789, "rewards/confidence_one_or_zero": 0.00166015625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.33313034772872924, "rewards/smce": -0.029536668956279755, "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": 0.0005859375, "completions/max_length": 1536.0, "completions/max_terminated_length": 409.8, "completions/mean_length": 118.90859375, "completions/mean_terminated_length": 118.07798919677734, "completions/min_length": 56.8, "completions/min_terminated_length": 56.8, "epoch": 0.24, "grad_norm": 0.0020928089506924152, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 255167868.0, "reward": 1.0354205131530763, "reward_std": 0.1417662888765335, "rewards/accuracy_reward": 0.3701171875, "rewards/brier_reward": 0.7690717101097106, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.32316468358039857, "rewards/smce": -0.06737189423292875, "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": 862.4, "completions/max_terminated_length": 383.6, "completions/mean_length": 116.2146484375, "completions/mean_terminated_length": 115.93745422363281, "completions/min_length": 56.0, "completions/min_terminated_length": 56.0, "epoch": 0.256, "grad_norm": 0.0021475371904671192, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 271412722.0, "reward": 1.039701771736145, "reward_std": 0.1479529157280922, "rewards/accuracy_reward": 0.3443359375, "rewards/brier_reward": 0.7795972108840943, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.331466943025589, "rewards/smce": -0.04423721497878432, "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": 804.6, "completions/max_terminated_length": 404.6, "completions/mean_length": 118.06962890625, "completions/mean_terminated_length": 117.7927017211914, "completions/min_length": 59.0, "completions/min_terminated_length": 59.0, "epoch": 0.272, "grad_norm": 0.0012407461181282997, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 287587451.0, "reward": 1.0504082918167115, "reward_std": 0.10368214398622513, "rewards/accuracy_reward": 0.36962890625, "rewards/brier_reward": 0.7729404926300049, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3405745208263397, "rewards/smce": -0.04136282093822956, "step": 85 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 477.0, "completions/max_terminated_length": 477.0, "completions/mean_length": 119.33857421875, "completions/mean_terminated_length": 119.33857421875, "completions/min_length": 57.8, "completions/min_terminated_length": 57.8, "epoch": 0.288, "grad_norm": 0.001116760540753603, "learning_rate": 1e-06, "loss": -0.0007, "num_tokens": 303767654.0, "reward": 1.0494994163513183, "reward_std": 0.13718811124563218, "rewards/accuracy_reward": 0.365625, "rewards/brier_reward": 0.7756176590919495, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.38375293016433715, "rewards/smce": -0.04204930327832699, "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": 627.4, "completions/max_terminated_length": 377.4, "completions/mean_length": 127.903125, "completions/mean_terminated_length": 127.76566162109376, "completions/min_length": 61.2, "completions/min_terminated_length": 61.2, "epoch": 0.304, "grad_norm": 0.002396478084847331, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 320007334.0, "reward": 1.0447796821594237, "reward_std": 0.22138679772615433, "rewards/accuracy_reward": 0.36904296875, "rewards/brier_reward": 0.7709169507026672, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.41757816076278687, "rewards/smce": -0.05001077130436897, "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": 830.0, "completions/max_terminated_length": 377.4, "completions/mean_length": 137.03095703125, "completions/mean_terminated_length": 136.7580078125, "completions/min_length": 59.8, "completions/min_terminated_length": 59.8, "epoch": 0.32, "grad_norm": 0.0017995622474700212, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 336499235.0, "reward": 1.0546803951263428, "reward_std": 0.13031208738684655, "rewards/accuracy_reward": 0.39091796875, "rewards/brier_reward": 0.7717086434364319, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.40168737173080443, "rewards/smce": -0.05287594236433506, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 249.0, "eval_completions/max_terminated_length": 249.0, "eval_completions/mean_length": 141.00047302246094, "eval_completions/mean_terminated_length": 141.00047302246094, "eval_completions/min_length": 71.5, "eval_completions/min_terminated_length": 71.5, "eval_loss": 0.0, "eval_num_tokens": 336499235.0, "eval_reward": 1.008398711681366, "eval_reward_std": 0.4021185487508774, "eval_rewards/accuracy_reward": 0.3046875, "eval_rewards/brier_reward": 0.7937804758548737, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.38554689288139343, "eval_rewards/smce": -0.08167129755020142, "eval_runtime": 9.8627, "eval_samples_per_second": 50.696, "eval_steps_per_second": 0.203, "step": 100 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 625.2, "completions/max_terminated_length": 418.2, "completions/mean_length": 143.5283203125, "completions/mean_terminated_length": 143.39237060546876, "completions/min_length": 62.2, "completions/min_terminated_length": 62.2, "epoch": 0.336, "grad_norm": 0.0010556376073509455, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 352691397.0, "reward": 1.0556965589523315, "reward_std": 0.156609907746315, "rewards/accuracy_reward": 0.39580078125, "rewards/brier_reward": 0.7704579830169678, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.38710381984710696, "rewards/smce": -0.05476872548460961, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 913.0, "completions/max_terminated_length": 505.8, "completions/mean_length": 152.359765625, "completions/mean_terminated_length": 152.0894989013672, "completions/min_length": 68.8, "completions/min_terminated_length": 68.8, "epoch": 0.352, "grad_norm": 0.0008620854932814837, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 369511977.0, "reward": 1.053104543685913, "reward_std": 0.12423949837684631, "rewards/accuracy_reward": 0.34990234375, "rewards/brier_reward": 0.7807921409606934, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3569404065608978, "rewards/smce": -0.024290787056088448, "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": 0.00029296875, "completions/max_length": 1111.4, "completions/max_terminated_length": 436.0, "completions/mean_length": 158.96328125, "completions/mean_terminated_length": 158.55966796875, "completions/min_length": 68.4, "completions/min_terminated_length": 68.4, "epoch": 0.368, "grad_norm": 0.0018711697775870562, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 386205233.0, "reward": 1.0552223086357118, "reward_std": 0.15189102590084075, "rewards/accuracy_reward": 0.3724609375, "rewards/brier_reward": 0.781423008441925, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.377591472864151, "rewards/smce": -0.042951859347522256, "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": 927.2, "completions/max_terminated_length": 491.4, "completions/mean_length": 164.28662109375, "completions/mean_terminated_length": 164.01817932128907, "completions/min_length": 70.0, "completions/min_terminated_length": 70.0, "epoch": 0.384, "grad_norm": 0.0008394431788474321, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 402744040.0, "reward": 1.070815658569336, "reward_std": 0.13146054446697236, "rewards/accuracy_reward": 0.3994140625, "rewards/brier_reward": 0.7760649085044861, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.39715608954429626, "rewards/smce": -0.033653135597705844, "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": 0.00029296875, "completions/max_length": 975.0, "completions/max_terminated_length": 543.0, "completions/mean_length": 166.998046875, "completions/mean_terminated_length": 166.59744262695312, "completions/min_length": 73.8, "completions/min_terminated_length": 73.8, "epoch": 0.4, "grad_norm": 0.0017887396970763803, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 419490548.0, "reward": 1.0607719659805297, "reward_std": 0.19124504327774047, "rewards/accuracy_reward": 0.37705078125, "rewards/brier_reward": 0.7800614953041076, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4110954165458679, "rewards/smce": -0.03517847005277872, "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": 0.00048828125, "completions/max_length": 1126.4, "completions/max_terminated_length": 536.4, "completions/mean_length": 168.71416015625, "completions/mean_terminated_length": 168.04581298828126, "completions/min_length": 73.8, "completions/min_terminated_length": 73.8, "epoch": 0.416, "grad_norm": 0.0007728107739239931, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 436099365.0, "reward": 1.0683367729187012, "reward_std": 0.10274968966841698, "rewards/accuracy_reward": 0.38662109375, "rewards/brier_reward": 0.7777512311935425, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.3897749066352844, "rewards/smce": -0.027113689482212065, "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": 0.000390625, "completions/max_length": 1134.6, "completions/max_terminated_length": 493.4, "completions/mean_length": 170.27578125, "completions/mean_terminated_length": 169.74155883789064, "completions/min_length": 71.4, "completions/min_terminated_length": 71.4, "epoch": 0.432, "grad_norm": 0.000608608650509268, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 452857325.0, "reward": 1.0728872060775756, "reward_std": 0.11906236484646797, "rewards/accuracy_reward": 0.41728515625, "rewards/brier_reward": 0.77127103805542, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.4053739070892334, "rewards/smce": -0.04219664148986339, "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": 614.4, "completions/max_terminated_length": 391.6, "completions/mean_length": 173.2181640625, "completions/mean_terminated_length": 173.08515014648438, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.448, "grad_norm": 0.0007781169260852039, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 469583879.0, "reward": 1.0613349199295044, "reward_std": 0.19281143099069595, "rewards/accuracy_reward": 0.39814453125, "rewards/brier_reward": 0.7739669680595398, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4190573275089264, "rewards/smce": -0.04924714155495167, "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": 0.0001953125, "completions/max_length": 716.8, "completions/max_terminated_length": 509.6, "completions/mean_length": 176.6033203125, "completions/mean_terminated_length": 176.33763732910157, "completions/min_length": 75.6, "completions/min_terminated_length": 75.6, "epoch": 0.464, "grad_norm": 0.001610193052329123, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 486563113.0, "reward": 1.0416291236877442, "reward_std": 0.17467037439346314, "rewards/accuracy_reward": 0.3498046875, "rewards/brier_reward": 0.7884636282920837, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3908169031143188, "rewards/smce": -0.05471789538860321, "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": 0.000390625, "completions/max_length": 1091.6, "completions/max_terminated_length": 458.8, "completions/mean_length": 173.646875, "completions/mean_terminated_length": 173.1146270751953, "completions/min_length": 77.8, "completions/min_terminated_length": 77.8, "epoch": 0.48, "grad_norm": 0.0008457148214802146, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 503389289.0, "reward": 1.0622626066207885, "reward_std": 0.0947317972779274, "rewards/accuracy_reward": 0.4201171875, "rewards/brier_reward": 0.7670642733573914, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.3603034496307373, "rewards/smce": -0.06207103803753853, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 346.0, "eval_completions/max_terminated_length": 346.0, "eval_completions/mean_length": 174.4136734008789, "eval_completions/mean_terminated_length": 174.4136734008789, "eval_completions/min_length": 79.5, "eval_completions/min_terminated_length": 79.5, "eval_loss": 0.0, "eval_num_tokens": 503389289.0, "eval_reward": 1.0325101017951965, "eval_reward_std": 0.2338552139699459, "eval_rewards/accuracy_reward": 0.322265625, "eval_rewards/brier_reward": 0.7960498034954071, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.3545898497104645, "eval_rewards/smce": -0.05329589918255806, "eval_runtime": 11.9083, "eval_samples_per_second": 41.987, "eval_steps_per_second": 0.168, "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": 952.4, "completions/max_terminated_length": 502.4, "completions/mean_length": 171.3103515625, "completions/mean_terminated_length": 171.0440460205078, "completions/min_length": 75.6, "completions/min_terminated_length": 75.6, "epoch": 0.496, "grad_norm": 0.0007915415917523205, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 520451347.0, "reward": 1.0678793430328368, "reward_std": 0.06508489176630974, "rewards/accuracy_reward": 0.405859375, "rewards/brier_reward": 0.7695850133895874, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3678733766078949, "rewards/smce": -0.03929574340581894, "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": 679.6, "completions/max_terminated_length": 461.0, "completions/mean_length": 168.5634765625, "completions/mean_terminated_length": 168.43018493652343, "completions/min_length": 66.6, "completions/min_terminated_length": 66.6, "epoch": 0.512, "grad_norm": 0.001805165666155517, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 537323101.0, "reward": 1.0813367128372193, "reward_std": 0.12635424956679345, "rewards/accuracy_reward": 0.42490234375, "rewards/brier_reward": 0.7725860834121704, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.42473828196525576, "rewards/smce": -0.034718182310462, "step": 160 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 899.2, "completions/max_terminated_length": 467.4, "completions/mean_length": 168.34658203125, "completions/mean_terminated_length": 168.0794708251953, "completions/min_length": 75.2, "completions/min_terminated_length": 75.2, "epoch": 0.528, "grad_norm": 0.0018557666335254908, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 554076506.0, "reward": 1.0747482538223267, "reward_std": 0.21446263790130615, "rewards/accuracy_reward": 0.4189453125, "rewards/brier_reward": 0.7756845116615295, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4623222589492798, "rewards/smce": -0.04493898786604404, "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": 0.00048828125, "completions/max_length": 1087.8, "completions/max_terminated_length": 410.4, "completions/mean_length": 167.5201171875, "completions/mean_terminated_length": 166.85220031738282, "completions/min_length": 71.0, "completions/min_terminated_length": 71.0, "epoch": 0.544, "grad_norm": 0.0006599732441827655, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 570955496.0, "reward": 1.0958893299102783, "reward_std": 0.132136981934309, "rewards/accuracy_reward": 0.45107421875, "rewards/brier_reward": 0.7667140364646912, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.4455615162849426, "rewards/smce": -0.025326891243457793, "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": 0.000390625, "completions/max_length": 863.8, "completions/max_terminated_length": 435.8, "completions/mean_length": 169.76455078125, "completions/mean_terminated_length": 169.23124084472656, "completions/min_length": 75.6, "completions/min_terminated_length": 75.6, "epoch": 0.56, "grad_norm": 0.0016425641952082515, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 587515293.0, "reward": 1.0665081977844237, "reward_std": 0.18351908326148986, "rewards/accuracy_reward": 0.39267578125, "rewards/brier_reward": 0.7788697123527527, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.423252534866333, "rewards/smce": -0.038139309734106064, "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": 9.765625e-05, "completions/max_length": 866.4, "completions/max_terminated_length": 654.8, "completions/mean_length": 169.991796875, "completions/mean_terminated_length": 169.85823974609374, "completions/min_length": 75.4, "completions/min_terminated_length": 75.4, "epoch": 0.576, "grad_norm": 0.0007072224980220199, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 604442633.0, "reward": 1.061542308330536, "reward_std": 0.07586740329861641, "rewards/accuracy_reward": 0.39912109375, "rewards/brier_reward": 0.7757653713226318, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.38846516609191895, "rewards/smce": -0.0517049060203135, "step": 180 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 490.0, "completions/max_terminated_length": 490.0, "completions/mean_length": 169.1283203125, "completions/mean_terminated_length": 169.1283203125, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.592, "grad_norm": 0.0006896816776134074, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 621342219.0, "reward": 1.0860573053359985, "reward_std": 0.07837061062455178, "rewards/accuracy_reward": 0.42060546875, "rewards/brier_reward": 0.775849437713623, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4021332383155823, "rewards/smce": -0.024243370443582535, "step": 185 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 855.4, "completions/max_terminated_length": 419.8, "completions/mean_length": 169.9259765625, "completions/mean_terminated_length": 169.65957641601562, "completions/min_length": 66.6, "completions/min_terminated_length": 66.6, "epoch": 0.608, "grad_norm": 0.0015429784543812275, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 638081749.0, "reward": 1.078372049331665, "reward_std": 0.14375375732779502, "rewards/accuracy_reward": 0.41201171875, "rewards/brier_reward": 0.7825367450714111, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4279746115207672, "rewards/smce": -0.037512218952178954, "step": 190 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 737.8, "completions/max_terminated_length": 545.6, "completions/mean_length": 171.880078125, "completions/mean_terminated_length": 171.74692993164064, "completions/min_length": 79.0, "completions/min_terminated_length": 79.0, "epoch": 0.624, "grad_norm": 0.001811556750908494, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 655185705.0, "reward": 1.0680161714553833, "reward_std": 0.1930210031569004, "rewards/accuracy_reward": 0.40322265625, "rewards/brier_reward": 0.7761505842208862, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44433175325393676, "rewards/smce": -0.04324411116540432, "step": 195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 867.0, "completions/max_terminated_length": 457.2, "completions/mean_length": 172.5982421875, "completions/mean_terminated_length": 172.331591796875, "completions/min_length": 77.6, "completions/min_terminated_length": 77.6, "epoch": 0.64, "grad_norm": 0.0007256731623783708, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 672295799.0, "reward": 1.0860810041427613, "reward_std": 0.12469884678721428, "rewards/accuracy_reward": 0.45810546875, "rewards/brier_reward": 0.7743456363677979, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4450886070728302, "rewards/smce": -0.06009464208036661, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 317.5, "eval_completions/max_terminated_length": 317.5, "eval_completions/mean_length": 174.84073638916016, "eval_completions/mean_terminated_length": 174.84073638916016, "eval_completions/min_length": 86.5, "eval_completions/min_terminated_length": 86.5, "eval_loss": 0.0, "eval_num_tokens": 672295799.0, "eval_reward": 1.032403826713562, "eval_reward_std": 0.44501954317092896, "eval_rewards/accuracy_reward": 0.359375, "eval_rewards/brier_reward": 0.7793614268302917, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4321940243244171, "eval_rewards/smce": -0.07392958551645279, "eval_runtime": 11.2627, "eval_samples_per_second": 44.394, "eval_steps_per_second": 0.178, "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": 672.4, "completions/max_terminated_length": 443.6, "completions/mean_length": 171.9009765625, "completions/mean_terminated_length": 171.76771850585936, "completions/min_length": 72.6, "completions/min_terminated_length": 72.6, "epoch": 0.656, "grad_norm": 0.0017715675057843328, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 688912609.0, "reward": 1.0504565954208374, "reward_std": 0.1744945839047432, "rewards/accuracy_reward": 0.3833984375, "rewards/brier_reward": 0.7761458277702331, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.43813673257827757, "rewards/smce": -0.0585342787206173, "step": 205 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 858.0, "completions/max_terminated_length": 403.2, "completions/mean_length": 169.2025390625, "completions/mean_terminated_length": 168.9353759765625, "completions/min_length": 77.6, "completions/min_terminated_length": 77.6, "epoch": 0.672, "grad_norm": 0.0013544198591262102, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 705558683.0, "reward": 1.0809198856353759, "reward_std": 0.07898051738739013, "rewards/accuracy_reward": 0.42177734375, "rewards/brier_reward": 0.7770021557807922, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.39942772388458253, "rewards/smce": -0.03674512431025505, "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": 930.6, "completions/max_terminated_length": 487.6, "completions/mean_length": 165.88359375, "completions/mean_terminated_length": 165.615380859375, "completions/min_length": 72.6, "completions/min_terminated_length": 72.6, "epoch": 0.688, "grad_norm": 0.0007912909495644271, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 722211251.0, "reward": 1.0810879468917847, "reward_std": 0.10564249902963638, "rewards/accuracy_reward": 0.42451171875, "rewards/brier_reward": 0.7782056927680969, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4082004904747009, "rewards/smce": -0.040249383077025414, "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": 637.6, "completions/max_terminated_length": 441.6, "completions/mean_length": 163.289453125, "completions/mean_terminated_length": 163.1553527832031, "completions/min_length": 74.8, "completions/min_terminated_length": 74.8, "epoch": 0.704, "grad_norm": 0.0007990578305907547, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 738749479.0, "reward": 1.1020585775375367, "reward_std": 0.05479091256856918, "rewards/accuracy_reward": 0.45537109375, "rewards/brier_reward": 0.7723845362663269, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.433248370885849, "rewards/smce": -0.023541683424264194, "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": 674.6, "completions/max_terminated_length": 523.8, "completions/mean_length": 160.61611328125, "completions/mean_terminated_length": 160.48172607421876, "completions/min_length": 76.0, "completions/min_terminated_length": 76.0, "epoch": 0.72, "grad_norm": 0.001754553522914648, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 755404044.0, "reward": 1.097388005256653, "reward_std": 0.16937963739037515, "rewards/accuracy_reward": 0.46162109375, "rewards/brier_reward": 0.7779244422912598, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.49155274629592893, "rewards/smce": -0.04467286169528961, "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": 897.2, "completions/max_terminated_length": 518.6, "completions/mean_length": 159.86923828125, "completions/mean_terminated_length": 159.6005859375, "completions/min_length": 74.8, "completions/min_terminated_length": 74.8, "epoch": 0.736, "grad_norm": 0.0007495178724639118, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 771980689.0, "reward": 1.0950091004371643, "reward_std": 0.15115555971860886, "rewards/accuracy_reward": 0.46552734375, "rewards/brier_reward": 0.7746174335479736, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.5008993089199066, "rewards/smce": -0.04963929690420628, "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": 610.4, "completions/max_terminated_length": 379.0, "completions/mean_length": 164.11552734375, "completions/mean_terminated_length": 163.98125305175782, "completions/min_length": 75.2, "completions/min_terminated_length": 75.2, "epoch": 0.752, "grad_norm": 0.0018389074830338359, "learning_rate": 1e-06, "loss": -0.0009, "num_tokens": 788888432.0, "reward": 1.0812556982040404, "reward_std": 0.18698384910821914, "rewards/accuracy_reward": 0.44228515625, "rewards/brier_reward": 0.7731466650962829, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4763873994350433, "rewards/smce": -0.05282366499304771, "step": 235 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 868.0, "completions/max_terminated_length": 417.2, "completions/mean_length": 170.9630859375, "completions/mean_terminated_length": 170.5629150390625, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.768, "grad_norm": 0.0017652820097282529, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 805571798.0, "reward": 1.1029137134552003, "reward_std": 0.1058156132698059, "rewards/accuracy_reward": 0.4453125, "rewards/brier_reward": 0.7754810214042663, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4427620470523834, "rewards/smce": -0.014673935808241367, "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": 481.2, "completions/max_terminated_length": 481.2, "completions/mean_length": 175.53525390625, "completions/mean_terminated_length": 175.53525390625, "completions/min_length": 79.8, "completions/min_terminated_length": 79.8, "epoch": 0.784, "grad_norm": 0.0016631303587928414, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 822543647.0, "reward": 1.0799616813659667, "reward_std": 0.08547558933496475, "rewards/accuracy_reward": 0.45078125, "rewards/brier_reward": 0.7736279129981994, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4300781190395355, "rewards/smce": -0.06448663324117661, "step": 245 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 505.2, "completions/max_terminated_length": 505.2, "completions/mean_length": 185.081640625, "completions/mean_terminated_length": 185.081640625, "completions/min_length": 84.4, "completions/min_terminated_length": 84.4, "epoch": 0.8, "grad_norm": 0.0006091457325965166, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 839449443.0, "reward": 1.1104381084442139, "reward_std": 0.07731564790010452, "rewards/accuracy_reward": 0.4669921875, "rewards/brier_reward": 0.7791252136230469, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.45222982168197634, "rewards/smce": -0.02504670396447182, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 360.0, "eval_completions/max_terminated_length": 360.0, "eval_completions/mean_length": 191.81483459472656, "eval_completions/mean_terminated_length": 191.81483459472656, "eval_completions/min_length": 106.5, "eval_completions/min_terminated_length": 106.5, "eval_loss": 0.0, "eval_num_tokens": 839449443.0, "eval_reward": 1.0259922742843628, "eval_reward_std": 0.450839638710022, "eval_rewards/accuracy_reward": 0.36328125, "eval_rewards/brier_reward": 0.7874349653720856, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 0.998046875, "eval_rewards/mean_confidence_reward": 0.4586132913827896, "eval_rewards/smce": -0.0967794917523861, "eval_runtime": 12.4313, "eval_samples_per_second": 40.221, "eval_steps_per_second": 0.161, "step": 250 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 461.4, "completions/max_terminated_length": 461.4, "completions/mean_length": 188.6380859375, "completions/mean_terminated_length": 188.6380859375, "completions/min_length": 87.4, "completions/min_terminated_length": 87.4, "epoch": 0.816, "grad_norm": 0.0016499223420396447, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 856480265.0, "reward": 1.1105870723724365, "reward_std": 0.183387403935194, "rewards/accuracy_reward": 0.46494140625, "rewards/brier_reward": 0.7762531995773315, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.48135270476341246, "rewards/smce": -0.02002142835408449, "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": 823.0, "completions/max_terminated_length": 634.0, "completions/mean_length": 202.40654296875, "completions/mean_terminated_length": 202.27649841308593, "completions/min_length": 97.2, "completions/min_terminated_length": 97.2, "epoch": 0.832, "grad_norm": 0.001645341981202364, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 873561260.0, "reward": 1.1026814222335815, "reward_std": 0.1493891693651676, "rewards/accuracy_reward": 0.46220703125, "rewards/brier_reward": 0.7856576681137085, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4738030254840851, "rewards/smce": -0.04240508079528808, "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": 1017.2, "completions/max_terminated_length": 619.2, "completions/mean_length": 212.58662109375, "completions/mean_terminated_length": 212.3273681640625, "completions/min_length": 100.2, "completions/min_terminated_length": 100.2, "epoch": 0.848, "grad_norm": 0.0004959266516380012, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 890752515.0, "reward": 1.0983349561691285, "reward_std": 0.17799243032932283, "rewards/accuracy_reward": 0.43671875, "rewards/brier_reward": 0.7796290874481201, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4496878385543823, "rewards/smce": -0.019385896250605585, "step": 265 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1124.6, "completions/max_terminated_length": 591.0, "completions/mean_length": 222.808203125, "completions/mean_terminated_length": 222.29569091796876, "completions/min_length": 95.8, "completions/min_terminated_length": 95.8, "epoch": 0.864, "grad_norm": 0.0007305259350687265, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 908020887.0, "reward": 1.1078871726989745, "reward_std": 0.06039171889424324, "rewards/accuracy_reward": 0.48857421875, "rewards/brier_reward": 0.7802868604660034, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.43827933073043823, "rewards/smce": -0.05269692949950695, "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": 0.0005859375, "completions/max_length": 1358.4, "completions/max_terminated_length": 706.8, "completions/mean_length": 227.19384765625, "completions/mean_terminated_length": 226.42540893554687, "completions/min_length": 103.2, "completions/min_terminated_length": 103.2, "epoch": 0.88, "grad_norm": 0.0006738873780705035, "learning_rate": 1e-06, "loss": 0.0023, "num_tokens": 925494424.0, "reward": 1.0921721696853637, "reward_std": 0.11363763809204101, "rewards/accuracy_reward": 0.43515625, "rewards/brier_reward": 0.7936494588851929, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.99921875, "rewards/mean_confidence_reward": 0.4402643144130707, "rewards/smce": -0.043680974096059796, "step": 275 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 710.2, "completions/max_terminated_length": 710.2, "completions/mean_length": 230.319921875, "completions/mean_terminated_length": 230.319921875, "completions/min_length": 108.0, "completions/min_terminated_length": 108.0, "epoch": 0.896, "grad_norm": 0.0014769054250791669, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 942963748.0, "reward": 1.090329909324646, "reward_std": 0.1920260950922966, "rewards/accuracy_reward": 0.44501953125, "rewards/brier_reward": 0.7832257986068726, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47868783473968507, "rewards/smce": -0.04748881347477436, "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": 0.00029296875, "completions/max_length": 1057.2, "completions/max_terminated_length": 777.2, "completions/mean_length": 235.7724609375, "completions/mean_terminated_length": 235.3920684814453, "completions/min_length": 104.0, "completions/min_terminated_length": 104.0, "epoch": 0.912, "grad_norm": 0.001404157723300159, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 960429354.0, "reward": 1.0865304470062256, "reward_std": 0.1757066637277603, "rewards/accuracy_reward": 0.44658203125, "rewards/brier_reward": 0.7791101217269898, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.467769855260849, "rewards/smce": -0.05224150661379099, "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": 593.2, "completions/max_terminated_length": 593.2, "completions/mean_length": 233.7447265625, "completions/mean_terminated_length": 233.7447265625, "completions/min_length": 112.4, "completions/min_terminated_length": 112.4, "epoch": 0.928, "grad_norm": 0.0005844378028996289, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 977849716.0, "reward": 1.1005787611007691, "reward_std": 0.09420083612203597, "rewards/accuracy_reward": 0.4736328125, "rewards/brier_reward": 0.7723412156105042, "rewards/confidence_one_or_zero": 0.00048828125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4323707938194275, "rewards/smce": -0.04481726624071598, "step": 290 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 1024.2, "completions/max_terminated_length": 659.2, "completions/mean_length": 238.26455078125, "completions/mean_terminated_length": 238.01092834472655, "completions/min_length": 105.8, "completions/min_terminated_length": 105.8, "epoch": 0.944, "grad_norm": 0.0006636297912336886, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 995264969.0, "reward": 1.0823294997215271, "reward_std": 0.1275633230805397, "rewards/accuracy_reward": 0.45625, "rewards/brier_reward": 0.7816440343856812, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4385166108608246, "rewards/smce": -0.07294286265969277, "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": 652.2, "completions/max_terminated_length": 652.2, "completions/mean_length": 240.97216796875, "completions/mean_terminated_length": 240.97216796875, "completions/min_length": 109.4, "completions/min_terminated_length": 109.4, "epoch": 0.96, "grad_norm": 0.0013080838834866881, "learning_rate": 1e-06, "loss": -0.0004, "num_tokens": 1012672844.0, "reward": 1.0975855112075805, "reward_std": 0.16411191448569298, "rewards/accuracy_reward": 0.43798828125, "rewards/brier_reward": 0.7877278208732605, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.45487695932388306, "rewards/smce": -0.03054591305553913, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 566.0, "eval_completions/max_terminated_length": 566.0, "eval_completions/mean_length": 250.1463165283203, "eval_completions/mean_terminated_length": 250.1463165283203, "eval_completions/min_length": 139.0, "eval_completions/min_terminated_length": 139.0, "eval_loss": 0.0, "eval_num_tokens": 1012672844.0, "eval_reward": 1.0451514720916748, "eval_reward_std": 0.4478505998849869, "eval_rewards/accuracy_reward": 0.376953125, "eval_rewards/brier_reward": 0.7926013767719269, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4547460973262787, "eval_rewards/smce": -0.0792523492127657, "eval_runtime": 16.3848, "eval_samples_per_second": 30.516, "eval_steps_per_second": 0.122, "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": 0.00048828125, "completions/max_length": 1536.0, "completions/max_terminated_length": 680.8, "completions/mean_length": 248.82431640625, "completions/mean_terminated_length": 248.19550476074218, "completions/min_length": 108.4, "completions/min_terminated_length": 108.4, "epoch": 0.976, "grad_norm": 0.0005834837211295962, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 1030081925.0, "reward": 1.1110716819763184, "reward_std": 0.10907015204429626, "rewards/accuracy_reward": 0.45869140625, "rewards/brier_reward": 0.7846518516540527, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.45255274176597593, "rewards/smce": -0.02071247100830078, "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": 0.0005859375, "completions/max_length": 1159.2, "completions/max_terminated_length": 645.4, "completions/mean_length": 248.54375, "completions/mean_terminated_length": 247.7925231933594, "completions/min_length": 115.6, "completions/min_terminated_length": 115.6, "epoch": 0.992, "grad_norm": 0.0006400683778338134, "learning_rate": 1e-06, "loss": 0.0019, "num_tokens": 1047755493.0, "reward": 1.0945802688598634, "reward_std": 0.1581621140241623, "rewards/accuracy_reward": 0.44541015625, "rewards/brier_reward": 0.7835812568664551, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.4432067096233368, "rewards/smce": -0.039245741069316865, "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": 0.00031887755102039117, "completions/max_length": 1156.0, "completions/max_terminated_length": 673.5, "completions/mean_length": 252.14827728271484, "completions/mean_terminated_length": 251.73896026611328, "completions/min_length": 110.5, "completions/min_terminated_length": 110.5, "epoch": 0.9984, "num_tokens": 1054795355.0, "reward": 1.0936598181724548, "reward_std": 0.13521113619208336, "rewards/accuracy_reward": 0.45751953125, "rewards/brier_reward": 0.7825449705123901, "rewards/confidence_one_or_zero": 0.000244140625, "rewards/format_reward": 0.999755859375, "rewards/mean_confidence_reward": 0.46235597133636475, "rewards/smce": -0.05250175856053829, "step": 312, "total_flos": 0.0, "train_loss": 0.002977700178095032, "train_runtime": 43698.358, "train_samples_per_second": 0.458, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1054795355, "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 }