{ "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": 0.11220703125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1528.6, "completions/mean_length": 466.6302734375, "completions/mean_terminated_length": 331.60613403320315, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.0026785461232066154, "learning_rate": 3.1249999999999997e-07, "loss": 0.0375, "num_tokens": 19622326.0, "reward": 0.09751158058643342, "reward_std": 0.4065922200679779, "rewards/accuracy_reward": 0.06201171875, "rewards/brier_reward": 0.10872447341680527, "rewards/confidence_one_or_zero": 0.41396484375, "rewards/format_reward": 0.259375, "rewards/mean_confidence_reward": 0.532010293006897, "rewards/smce": -0.47017982006073, "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": 0.10126953125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1532.6, "completions/mean_length": 446.04921875, "completions/mean_terminated_length": 323.296875, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.00424788985401392, "learning_rate": 6.249999999999999e-07, "loss": 0.0371, "num_tokens": 39290222.0, "reward": 0.11004470586776734, "reward_std": 0.4158190131187439, "rewards/accuracy_reward": 0.06279296875, "rewards/brier_reward": 0.11651045382022858, "rewards/confidence_one_or_zero": 0.43310546875, "rewards/format_reward": 0.28916015625, "rewards/mean_confidence_reward": 0.559346604347229, "rewards/smce": -0.49675229787826536, "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": 0.05732421875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1501.2, "completions/mean_length": 317.6384765625, "completions/mean_terminated_length": 243.9247039794922, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.048, "grad_norm": 0.004304820206016302, "learning_rate": 9.374999999999999e-07, "loss": 0.0425, "num_tokens": 57591576.0, "reward": 0.2621927112340927, "reward_std": 0.4705701291561127, "rewards/accuracy_reward": 0.10595703125, "rewards/brier_reward": 0.19967454373836518, "rewards/confidence_one_or_zero": 0.5154296875, "rewards/format_reward": 0.50380859375, "rewards/mean_confidence_reward": 0.6753389239311218, "rewards/smce": -0.5701142072677612, "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": 0.02685546875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1411.4, "completions/mean_length": 222.1150390625, "completions/mean_terminated_length": 186.05282897949218, "completions/min_length": 8.8, "completions/min_terminated_length": 8.8, "epoch": 0.064, "grad_norm": 0.003465007059276104, "learning_rate": 1e-06, "loss": 0.0317, "num_tokens": 74784434.0, "reward": 0.5405649662017822, "reward_std": 0.463888019323349, "rewards/accuracy_reward": 0.1904296875, "rewards/brier_reward": 0.35292364954948424, "rewards/confidence_one_or_zero": 0.53916015625, "rewards/format_reward": 0.8271484375, "rewards/mean_confidence_reward": 0.7663158893585205, "rewards/smce": -0.5787488460540772, "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": 0.012890625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1265.8, "completions/mean_length": 169.68251953125, "completions/mean_terminated_length": 151.84535522460936, "completions/min_length": 29.4, "completions/min_terminated_length": 29.4, "epoch": 0.08, "grad_norm": 0.003266093088313937, "learning_rate": 1e-06, "loss": 0.0177, "num_tokens": 91455135.0, "reward": 0.6993292093276977, "reward_std": 0.40993152260780336, "rewards/accuracy_reward": 0.2166015625, "rewards/brier_reward": 0.4692054033279419, "rewards/confidence_one_or_zero": 0.443359375, "rewards/format_reward": 0.96181640625, "rewards/mean_confidence_reward": 0.7073621034622193, "rewards/smce": -0.49793441891670226, "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": 0.0080078125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1338.0, "completions/mean_length": 152.65791015625, "completions/mean_terminated_length": 141.49590148925782, "completions/min_length": 34.2, "completions/min_terminated_length": 34.2, "epoch": 0.096, "grad_norm": 0.00212898850440979, "learning_rate": 1e-06, "loss": 0.0139, "num_tokens": 108062960.0, "reward": 0.8696157813072205, "reward_std": 0.3268528938293457, "rewards/accuracy_reward": 0.23359375, "rewards/brier_reward": 0.6494227528572083, "rewards/confidence_one_or_zero": 0.310546875, "rewards/format_reward": 0.9818359375, "rewards/mean_confidence_reward": 0.4570514678955078, "rewards/smce": -0.25124485343694686, "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": 0.002734375, "completions/max_length": 1536.0, "completions/max_terminated_length": 960.4, "completions/mean_length": 131.7365234375, "completions/mean_terminated_length": 127.88609161376954, "completions/min_length": 32.6, "completions/min_terminated_length": 32.6, "epoch": 0.112, "grad_norm": 0.0015331514878198504, "learning_rate": 1e-06, "loss": 0.0054, "num_tokens": 124521414.0, "reward": 0.9907561421394349, "reward_std": 0.1528225064277649, "rewards/accuracy_reward": 0.2646484375, "rewards/brier_reward": 0.7692283868789673, "rewards/confidence_one_or_zero": 0.3392578125, "rewards/format_reward": 0.9931640625, "rewards/mean_confidence_reward": 0.21056363880634307, "rewards/smce": -0.09105930775403977, "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": 664.8, "completions/mean_length": 125.91826171875, "completions/mean_terminated_length": 123.85120544433593, "completions/min_length": 40.6, "completions/min_terminated_length": 40.6, "epoch": 0.128, "grad_norm": 0.000771750696003437, "learning_rate": 1e-06, "loss": 0.002, "num_tokens": 140727489.0, "reward": 0.9993586897850036, "reward_std": 0.09848072528839111, "rewards/accuracy_reward": 0.2533203125, "rewards/brier_reward": 0.791638457775116, "rewards/confidence_one_or_zero": 0.256640625, "rewards/format_reward": 0.99677734375, "rewards/mean_confidence_reward": 0.18074644207954407, "rewards/smce": -0.08603915795683861, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 1335.4, "completions/max_terminated_length": 510.2, "completions/mean_length": 123.43896484375, "completions/mean_terminated_length": 122.61112060546876, "completions/min_length": 45.2, "completions/min_terminated_length": 45.2, "epoch": 0.144, "grad_norm": 0.0008707439992576838, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 156941936.0, "reward": 1.0172405004501344, "reward_std": 0.09546828716993332, "rewards/accuracy_reward": 0.2875, "rewards/brier_reward": 0.7847882270812988, "rewards/confidence_one_or_zero": 0.12060546875, "rewards/format_reward": 0.99716796875, "rewards/mean_confidence_reward": 0.2260705977678299, "rewards/smce": -0.06995189264416694, "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": 0.00078125, "completions/max_length": 1536.0, "completions/max_terminated_length": 543.6, "completions/mean_length": 127.2958984375, "completions/mean_terminated_length": 126.1939910888672, "completions/min_length": 55.4, "completions/min_terminated_length": 55.4, "epoch": 0.16, "grad_norm": 0.0015283646062016487, "learning_rate": 1e-06, "loss": 0.0019, "num_tokens": 173266374.0, "reward": 1.0233920097351075, "reward_std": 0.15360372960567475, "rewards/accuracy_reward": 0.2744140625, "rewards/brier_reward": 0.7956377387046814, "rewards/confidence_one_or_zero": 0.04130859375, "rewards/format_reward": 0.9982421875, "rewards/mean_confidence_reward": 0.2933915197849274, "rewards/smce": -0.043021409586071965, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.002049180327868827, "eval_completions/max_length": 909.5, "eval_completions/max_terminated_length": 406.0, "eval_completions/mean_length": 132.36570739746094, "eval_completions/mean_terminated_length": 129.48561096191406, "eval_completions/min_length": 62.5, "eval_completions/min_terminated_length": 62.5, "eval_loss": 0.0, "eval_num_tokens": 173266374.0, "eval_reward": 0.9988523125648499, "eval_reward_std": 0.2654736191034317, "eval_rewards/accuracy_reward": 0.232421875, "eval_rewards/brier_reward": 0.8115327954292297, "eval_rewards/confidence_one_or_zero": 0.01953125, "eval_rewards/format_reward": 0.998046875, "eval_rewards/mean_confidence_reward": 0.3196093738079071, "eval_rewards/smce": -0.08859512209892273, "eval_runtime": 21.2363, "eval_samples_per_second": 23.545, "eval_steps_per_second": 0.094, "step": 50 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 676.2, "completions/max_terminated_length": 482.0, "completions/mean_length": 126.12138671875, "completions/mean_terminated_length": 125.98388366699218, "completions/min_length": 53.6, "completions/min_terminated_length": 53.6, "epoch": 0.176, "grad_norm": 0.001403786358423531, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 189794977.0, "reward": 1.0303616762161254, "reward_std": 0.1711978167295456, "rewards/accuracy_reward": 0.28095703125, "rewards/brier_reward": 0.798363721370697, "rewards/confidence_one_or_zero": 0.02021484375, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.3105536162853241, "rewards/smce": -0.03524293936789036, "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": 0.0001953125, "completions/max_length": 938.8, "completions/max_terminated_length": 484.0, "completions/mean_length": 126.25869140625, "completions/mean_terminated_length": 125.98303680419922, "completions/min_length": 55.6, "completions/min_terminated_length": 55.6, "epoch": 0.192, "grad_norm": 0.0014057139633223414, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 205902682.0, "reward": 1.0450642347335815, "reward_std": 0.0979317307472229, "rewards/accuracy_reward": 0.33349609375, "rewards/brier_reward": 0.7794028878211975, "rewards/confidence_one_or_zero": 0.00712890625, "rewards/format_reward": 0.99912109375, "rewards/mean_confidence_reward": 0.3114392042160034, "rewards/smce": -0.04378440342843533, "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": 0.000390625, "completions/max_length": 1121.8, "completions/max_terminated_length": 461.8, "completions/mean_length": 129.217578125, "completions/mean_terminated_length": 128.668408203125, "completions/min_length": 56.4, "completions/min_terminated_length": 56.4, "epoch": 0.208, "grad_norm": 0.0007134112529456615, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 222258094.0, "reward": 1.0534921646118165, "reward_std": 0.09793903082609176, "rewards/accuracy_reward": 0.35361328125, "rewards/brier_reward": 0.7721356630325318, "rewards/confidence_one_or_zero": 0.0029296875, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.32707850337028505, "rewards/smce": -0.03557743579149246, "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": 806.2, "completions/max_terminated_length": 561.2, "completions/mean_length": 126.786328125, "completions/mean_terminated_length": 126.64864959716797, "completions/min_length": 56.8, "completions/min_terminated_length": 56.8, "epoch": 0.224, "grad_norm": 0.0014312885468825698, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 238709570.0, "reward": 1.0552095890045166, "reward_std": 0.11882918179035187, "rewards/accuracy_reward": 0.34658203125, "rewards/brier_reward": 0.7787379264831543, "rewards/confidence_one_or_zero": 0.0013671875, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3532362937927246, "rewards/smce": -0.029021762683987616, "step": 70 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 750.0, "completions/max_terminated_length": 540.4, "completions/mean_length": 128.212109375, "completions/mean_terminated_length": 128.07445831298827, "completions/min_length": 56.0, "completions/min_terminated_length": 56.0, "epoch": 0.24, "grad_norm": 0.0015375501243397593, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 255274142.0, "reward": 1.0612101793289184, "reward_std": 0.14273426681756973, "rewards/accuracy_reward": 0.39052734375, "rewards/brier_reward": 0.7600910544395447, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.3782422184944153, "rewards/smce": -0.055030280724167824, "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": 0.0, "completions/max_length": 389.4, "completions/max_terminated_length": 389.4, "completions/mean_length": 127.03037109375, "completions/mean_terminated_length": 127.03037109375, "completions/min_length": 57.2, "completions/min_terminated_length": 57.2, "epoch": 0.256, "grad_norm": 0.0014671719400212169, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 271629749.0, "reward": 1.0515930891036986, "reward_std": 0.1513766437768936, "rewards/accuracy_reward": 0.35771484375, "rewards/brier_reward": 0.7751455187797547, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.3833678364753723, "rewards/smce": -0.059154579788446425, "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": 9.765625e-05, "completions/max_length": 609.4, "completions/max_terminated_length": 382.0, "completions/mean_length": 133.298828125, "completions/mean_terminated_length": 133.16205444335938, "completions/min_length": 60.6, "completions/min_terminated_length": 60.6, "epoch": 0.272, "grad_norm": 0.0006801378331147134, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 287960425.0, "reward": 1.0686883926391602, "reward_std": 0.08673371747136116, "rewards/accuracy_reward": 0.3943359375, "rewards/brier_reward": 0.7641499638557434, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.36116114258766174, "rewards/smce": -0.041243164986372, "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": 601.4, "completions/max_terminated_length": 365.2, "completions/mean_length": 131.62919921875, "completions/mean_terminated_length": 131.4923095703125, "completions/min_length": 58.8, "completions/min_terminated_length": 58.8, "epoch": 0.288, "grad_norm": 0.0007481302018277347, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 304266484.0, "reward": 1.0655381202697753, "reward_std": 0.10698467046022415, "rewards/accuracy_reward": 0.38828125, "rewards/brier_reward": 0.7688133239746093, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3731468141078949, "rewards/smce": -0.05125699490308762, "step": 90 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 864.2, "completions/max_terminated_length": 418.8, "completions/mean_length": 132.52294921875, "completions/mean_terminated_length": 132.2486785888672, "completions/min_length": 58.2, "completions/min_terminated_length": 58.2, "epoch": 0.304, "grad_norm": 0.0016072518192231655, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 320553471.0, "reward": 1.072618842124939, "reward_std": 0.0957090824842453, "rewards/accuracy_reward": 0.39638671875, "rewards/brier_reward": 0.7660355925559997, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.39573111534118655, "rewards/smce": -0.03378484845161438, "step": 95 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 726.0, "completions/max_terminated_length": 483.8, "completions/mean_length": 134.47216796875, "completions/mean_terminated_length": 134.33522033691406, "completions/min_length": 61.0, "completions/min_terminated_length": 61.0, "epoch": 0.32, "grad_norm": 0.0012717916397377849, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 337019170.0, "reward": 1.0765431642532348, "reward_std": 0.12678306698799133, "rewards/accuracy_reward": 0.414453125, "rewards/brier_reward": 0.768175745010376, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.42739094495773317, "rewards/smce": -0.058696174249053004, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 261.0, "eval_completions/max_terminated_length": 261.0, "eval_completions/mean_length": 137.20244598388672, "eval_completions/mean_terminated_length": 137.20244598388672, "eval_completions/min_length": 68.5, "eval_completions/min_terminated_length": 68.5, "eval_loss": 0.0, "eval_num_tokens": 337019170.0, "eval_reward": 1.0082088112831116, "eval_reward_std": 0.3224063217639923, "eval_rewards/accuracy_reward": 0.3046875, "eval_rewards/brier_reward": 0.7768570184707642, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.43464845418930054, "eval_rewards/smce": -0.13025547564029694, "eval_runtime": 10.2855, "eval_samples_per_second": 48.612, "eval_steps_per_second": 0.194, "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": 871.0, "completions/max_terminated_length": 389.2, "completions/mean_length": 133.99482421875, "completions/mean_terminated_length": 133.72105102539064, "completions/min_length": 55.6, "completions/min_terminated_length": 55.6, "epoch": 0.336, "grad_norm": 0.0007206884911283851, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 353113709.0, "reward": 1.0716354370117187, "reward_std": 0.13324915170669555, "rewards/accuracy_reward": 0.40849609375, "rewards/brier_reward": 0.7654546618461608, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4270427763462067, "rewards/smce": -0.06077566109597683, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 681.4, "completions/max_terminated_length": 465.2, "completions/mean_length": 137.42919921875, "completions/mean_terminated_length": 137.29295349121094, "completions/min_length": 60.8, "completions/min_terminated_length": 60.8, "epoch": 0.352, "grad_norm": 0.0013820958556607366, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 369781400.0, "reward": 1.0609832763671876, "reward_std": 0.14019967019557952, "rewards/accuracy_reward": 0.36474609375, "rewards/brier_reward": 0.7737235069274903, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.39401292204856875, "rewards/smce": -0.032421684823930264, "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": 665.6, "completions/max_terminated_length": 442.0, "completions/mean_length": 142.323046875, "completions/mean_terminated_length": 142.18718872070312, "completions/min_length": 61.6, "completions/min_terminated_length": 61.6, "epoch": 0.368, "grad_norm": 0.0005504830041900277, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 386304260.0, "reward": 1.0715895414352417, "reward_std": 0.09788179844617843, "rewards/accuracy_reward": 0.38828125, "rewards/brier_reward": 0.7760221242904664, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.383696848154068, "rewards/smce": -0.04146903231739998, "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": 981.8, "completions/max_terminated_length": 523.0, "completions/mean_length": 149.26435546875, "completions/mean_terminated_length": 148.9933074951172, "completions/min_length": 61.8, "completions/min_terminated_length": 61.8, "epoch": 0.384, "grad_norm": 0.0007317579002119601, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 402689239.0, "reward": 1.082389736175537, "reward_std": 0.08156476095318795, "rewards/accuracy_reward": 0.41943359375, "rewards/brier_reward": 0.7709457755088807, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.379436719417572, "rewards/smce": -0.05081076994538307, "step": 120 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 418.0, "completions/max_terminated_length": 418.0, "completions/mean_length": 154.80625, "completions/mean_terminated_length": 154.80625, "completions/min_length": 64.4, "completions/min_terminated_length": 64.4, "epoch": 0.4, "grad_norm": 0.001232767477631569, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 419310903.0, "reward": 1.0757294654846192, "reward_std": 0.12325254082679749, "rewards/accuracy_reward": 0.39443359375, "rewards/brier_reward": 0.7765468120574951, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4125796616077423, "rewards/smce": -0.0388492189347744, "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": 0.00068359375, "completions/max_length": 1536.0, "completions/max_terminated_length": 475.6, "completions/mean_length": 161.99873046875, "completions/mean_terminated_length": 161.05875549316406, "completions/min_length": 59.6, "completions/min_terminated_length": 59.6, "epoch": 0.416, "grad_norm": 0.0006661260849796236, "learning_rate": 1e-06, "loss": 0.0016, "num_tokens": 435850954.0, "reward": 1.082474160194397, "reward_std": 0.1305944874882698, "rewards/accuracy_reward": 0.40322265625, "rewards/brier_reward": 0.7749996423721314, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9990234375, "rewards/mean_confidence_reward": 0.41062086820602417, "rewards/smce": -0.02459658682346344, "step": 130 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 668.6, "completions/max_terminated_length": 445.4, "completions/mean_length": 166.15087890625, "completions/mean_terminated_length": 166.01694946289064, "completions/min_length": 70.0, "completions/min_terminated_length": 70.0, "epoch": 0.432, "grad_norm": 0.0005438687512651086, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 452566675.0, "reward": 1.0882618188858033, "reward_std": 0.09735962152481079, "rewards/accuracy_reward": 0.4267578125, "rewards/brier_reward": 0.7717886209487915, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.41006878614425657, "rewards/smce": -0.043851888924837115, "step": 135 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 956.2, "completions/max_terminated_length": 521.0, "completions/mean_length": 171.17646484375, "completions/mean_terminated_length": 170.77630920410155, "completions/min_length": 73.4, "completions/min_terminated_length": 73.4, "epoch": 0.448, "grad_norm": 0.0005960149574093521, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 469272322.0, "reward": 1.085831379890442, "reward_std": 0.12048186510801315, "rewards/accuracy_reward": 0.41865234375, "rewards/brier_reward": 0.7699586391448975, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.41828733682632446, "rewards/smce": -0.03331212252378464, "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": 0.00029296875, "completions/max_length": 896.6, "completions/max_terminated_length": 487.2, "completions/mean_length": 175.76884765625, "completions/mean_terminated_length": 175.37088317871093, "completions/min_length": 73.8, "completions/min_terminated_length": 73.8, "epoch": 0.464, "grad_norm": 0.001222067279741168, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 486243011.0, "reward": 1.057556962966919, "reward_std": 0.1313529819250107, "rewards/accuracy_reward": 0.36416015625, "rewards/brier_reward": 0.7829479813575745, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.411750328540802, "rewards/smce": -0.06340402290225029, "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": 0.00029296875, "completions/max_length": 921.0, "completions/max_terminated_length": 477.6, "completions/mean_length": 175.8951171875, "completions/mean_terminated_length": 175.4965362548828, "completions/min_length": 76.4, "completions/min_terminated_length": 76.4, "epoch": 0.48, "grad_norm": 0.0006074010743759573, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 503092209.0, "reward": 1.083800721168518, "reward_std": 0.09710027799010276, "rewards/accuracy_reward": 0.42744140625, "rewards/brier_reward": 0.7678929805755615, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.3819667637348175, "rewards/smce": -0.05488146059215069, "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": 179.1521453857422, "eval_completions/mean_terminated_length": 179.1521453857422, "eval_completions/min_length": 83.5, "eval_completions/min_terminated_length": 83.5, "eval_loss": 0.0, "eval_num_tokens": 503092209.0, "eval_reward": 1.055518388748169, "eval_reward_std": 0.29327547550201416, "eval_rewards/accuracy_reward": 0.333984375, "eval_rewards/brier_reward": 0.793656051158905, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.3653808832168579, "eval_rewards/smce": -0.03320850059390068, "eval_runtime": 11.8844, "eval_samples_per_second": 42.072, "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": 0.0, "completions/max_length": 498.4, "completions/max_terminated_length": 498.4, "completions/mean_length": 175.06181640625, "completions/mean_terminated_length": 175.06181640625, "completions/min_length": 79.0, "completions/min_terminated_length": 79.0, "epoch": 0.496, "grad_norm": 0.0006104473723098636, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 520192682.0, "reward": 1.0852439403533936, "reward_std": 0.057844220846891406, "rewards/accuracy_reward": 0.4291015625, "rewards/brier_reward": 0.7675500392913819, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.3779485821723938, "rewards/smce": -0.05232905186712742, "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": 437.6, "completions/max_terminated_length": 437.6, "completions/mean_length": 174.67958984375, "completions/mean_terminated_length": 174.67958984375, "completions/min_length": 82.6, "completions/min_terminated_length": 82.6, "epoch": 0.512, "grad_norm": 0.0012512748362496495, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 537127065.0, "reward": 1.099705195426941, "reward_std": 0.1057943195104599, "rewards/accuracy_reward": 0.44609375, "rewards/brier_reward": 0.7742470264434814, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.42951337695121766, "rewards/smce": -0.04186231698840857, "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": 964.8, "completions/max_terminated_length": 511.0, "completions/mean_length": 176.96630859375, "completions/mean_terminated_length": 176.70090942382814, "completions/min_length": 81.6, "completions/min_terminated_length": 81.6, "epoch": 0.528, "grad_norm": 0.0014094150392338634, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 553968736.0, "reward": 1.089972233772278, "reward_std": 0.16099656820297242, "rewards/accuracy_reward": 0.4208984375, "rewards/brier_reward": 0.7824561715126037, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.46478373408317564, "rewards/smce": -0.04643128663301468, "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": 880.4, "completions/max_terminated_length": 468.6, "completions/mean_length": 174.324609375, "completions/mean_terminated_length": 173.92534790039062, "completions/min_length": 81.6, "completions/min_terminated_length": 81.6, "epoch": 0.544, "grad_norm": 0.0006505737546831369, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 570917404.0, "reward": 1.1095312356948852, "reward_std": 0.11838259994983673, "rewards/accuracy_reward": 0.46826171875, "rewards/brier_reward": 0.7663807153701783, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4589267909526825, "rewards/smce": -0.030575690418481828, "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": 0.00029296875, "completions/max_length": 897.6, "completions/max_terminated_length": 467.6, "completions/mean_length": 172.4845703125, "completions/mean_terminated_length": 172.0850830078125, "completions/min_length": 77.4, "completions/min_terminated_length": 77.4, "epoch": 0.56, "grad_norm": 0.0012473545502871275, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 587505054.0, "reward": 1.086290669441223, "reward_std": 0.14059744775295258, "rewards/accuracy_reward": 0.41142578125, "rewards/brier_reward": 0.7802677392959595, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.44028844833374026, "rewards/smce": -0.03764005638659, "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": 0.000390625, "completions/max_length": 1094.4, "completions/max_terminated_length": 501.2, "completions/mean_length": 173.42021484375, "completions/mean_terminated_length": 172.88742065429688, "completions/min_length": 80.0, "completions/min_terminated_length": 80.0, "epoch": 0.576, "grad_norm": 0.0005928562022745609, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 604467501.0, "reward": 1.0810141086578369, "reward_std": 0.08423209637403488, "rewards/accuracy_reward": 0.4126953125, "rewards/brier_reward": 0.777400267124176, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.3991813659667969, "rewards/smce": -0.05535511486232281, "step": 180 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 863.8, "completions/max_terminated_length": 490.0, "completions/mean_length": 172.092578125, "completions/mean_terminated_length": 171.82615966796874, "completions/min_length": 79.4, "completions/min_terminated_length": 79.4, "epoch": 0.592, "grad_norm": 0.0005724472575820982, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 621397441.0, "reward": 1.0966659307479858, "reward_std": 0.05616953819990158, "rewards/accuracy_reward": 0.43623046875, "rewards/brier_reward": 0.7744009375572205, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.40447003245353697, "rewards/smce": -0.03421012945473194, "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": 0.00029296875, "completions/max_length": 1109.8, "completions/max_terminated_length": 437.0, "completions/mean_length": 174.52080078125, "completions/mean_terminated_length": 174.12187805175782, "completions/min_length": 78.2, "completions/min_terminated_length": 78.2, "epoch": 0.608, "grad_norm": 0.0029141635168343782, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 638184022.0, "reward": 1.0967584371566772, "reward_std": 0.11957666501402855, "rewards/accuracy_reward": 0.42626953125, "rewards/brier_reward": 0.7853588700294495, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4422805607318878, "rewards/smce": -0.03563878741115332, "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": 694.8, "completions/max_terminated_length": 488.0, "completions/mean_length": 178.27265625, "completions/mean_terminated_length": 178.1399719238281, "completions/min_length": 83.6, "completions/min_terminated_length": 83.6, "epoch": 0.624, "grad_norm": 0.0013668371830135584, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 655353438.0, "reward": 1.0849785327911377, "reward_std": 0.14437882453203202, "rewards/accuracy_reward": 0.4138671875, "rewards/brier_reward": 0.7801495552062988, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.45922787189483644, "rewards/smce": -0.047925911750644444, "step": 195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1099.6, "completions/max_terminated_length": 477.2, "completions/mean_length": 176.486328125, "completions/mean_terminated_length": 176.0878173828125, "completions/min_length": 84.8, "completions/min_terminated_length": 84.8, "epoch": 0.64, "grad_norm": 0.0005942194839008152, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 672503346.0, "reward": 1.117413091659546, "reward_std": 0.09016603529453278, "rewards/accuracy_reward": 0.48681640625, "rewards/brier_reward": 0.7776980638504029, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.463961261510849, "rewards/smce": -0.05859697572886944, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 385.5, "eval_completions/max_terminated_length": 385.5, "eval_completions/mean_length": 179.0931396484375, "eval_completions/mean_terminated_length": 179.0931396484375, "eval_completions/min_length": 93.0, "eval_completions/min_terminated_length": 93.0, "eval_loss": 0.0, "eval_num_tokens": 672503346.0, "eval_reward": 1.0669192671775818, "eval_reward_std": 0.3463890254497528, "eval_rewards/accuracy_reward": 0.38671875, "eval_rewards/brier_reward": 0.7844113707542419, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4601171910762787, "eval_rewards/smce": -0.0745849683880806, "eval_runtime": 12.3788, "eval_samples_per_second": 40.392, "eval_steps_per_second": 0.162, "step": 200 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 455.2, "completions/max_terminated_length": 455.2, "completions/mean_length": 172.68984375, "completions/mean_terminated_length": 172.68984375, "completions/min_length": 78.0, "completions/min_terminated_length": 78.0, "epoch": 0.656, "grad_norm": 0.001317652058787644, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 689128234.0, "reward": 1.0755244255065919, "reward_std": 0.14942533075809478, "rewards/accuracy_reward": 0.40205078125, "rewards/brier_reward": 0.7797590136528015, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4621377050876617, "rewards/smce": -0.06132837850600481, "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": 650.0, "completions/max_terminated_length": 423.2, "completions/mean_length": 167.75009765625, "completions/mean_terminated_length": 167.48269653320312, "completions/min_length": 77.4, "completions/min_terminated_length": 77.4, "epoch": 0.672, "grad_norm": 0.0010439743055030704, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 705759435.0, "reward": 1.0993054628372192, "reward_std": 0.07033557817339897, "rewards/accuracy_reward": 0.43154296875, "rewards/brier_reward": 0.7837408065795899, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4168268620967865, "rewards/smce": -0.03276139870285988, "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": 426.0, "completions/max_terminated_length": 426.0, "completions/mean_length": 166.26279296875, "completions/mean_terminated_length": 166.26279296875, "completions/min_length": 79.4, "completions/min_terminated_length": 79.4, "epoch": 0.688, "grad_norm": 0.0007595060160383582, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 722415886.0, "reward": 1.0981730222702026, "reward_std": 0.09036669880151749, "rewards/accuracy_reward": 0.43310546875, "rewards/brier_reward": 0.7850585222244263, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4186152458190918, "rewards/smce": -0.04344206675887108, "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": 610.0, "completions/max_terminated_length": 379.4, "completions/mean_length": 166.392578125, "completions/mean_terminated_length": 166.25871887207032, "completions/min_length": 76.8, "completions/min_terminated_length": 76.8, "epoch": 0.704, "grad_norm": 0.0007093519088812172, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 738985890.0, "reward": 1.111887812614441, "reward_std": 0.07709287628531455, "rewards/accuracy_reward": 0.45732421875, "rewards/brier_reward": 0.7776798009872437, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4373369514942169, "rewards/smce": -0.02226333422586322, "step": 220 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 622.0, "completions/max_terminated_length": 393.8, "completions/mean_length": 169.74150390625, "completions/mean_terminated_length": 169.47537841796876, "completions/min_length": 77.6, "completions/min_terminated_length": 77.6, "epoch": 0.72, "grad_norm": 0.0014726045774295926, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 755733899.0, "reward": 1.1221982479095458, "reward_std": 0.11447475254535674, "rewards/accuracy_reward": 0.4787109375, "rewards/brier_reward": 0.7866963624954224, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4943776249885559, "rewards/smce": -0.041632843762636186, "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": 874.8, "completions/max_terminated_length": 444.8, "completions/mean_length": 176.96787109375, "completions/mean_terminated_length": 176.70293579101562, "completions/min_length": 82.6, "completions/min_terminated_length": 82.6, "epoch": 0.736, "grad_norm": 0.0012901662848889828, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 772485634.0, "reward": 1.1121197938919067, "reward_std": 0.14175231754779816, "rewards/accuracy_reward": 0.47734375, "rewards/brier_reward": 0.7755715608596802, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5298642635345459, "rewards/smce": -0.056962878257036206, "step": 230 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 444.0, "completions/max_terminated_length": 444.0, "completions/mean_length": 181.18896484375, "completions/mean_terminated_length": 181.18896484375, "completions/min_length": 88.4, "completions/min_terminated_length": 88.4, "epoch": 0.752, "grad_norm": 0.0012587190140038729, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 789568209.0, "reward": 1.1084522962570191, "reward_std": 0.13682172298431397, "rewards/accuracy_reward": 0.45869140625, "rewards/brier_reward": 0.7826623439788818, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4847588062286377, "rewards/smce": -0.04890004470944405, "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": 0.0001953125, "completions/max_length": 871.8, "completions/max_terminated_length": 433.6, "completions/mean_length": 184.03701171875, "completions/mean_terminated_length": 183.77239074707032, "completions/min_length": 85.0, "completions/min_terminated_length": 85.0, "epoch": 0.768, "grad_norm": 0.0006104763597249985, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 806385452.0, "reward": 1.1183516025543212, "reward_std": 0.07741678655147552, "rewards/accuracy_reward": 0.4650390625, "rewards/brier_reward": 0.7864425659179688, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4476728618144989, "rewards/smce": -0.02916794307529926, "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": 472.0, "completions/max_terminated_length": 472.0, "completions/mean_length": 185.95849609375, "completions/mean_terminated_length": 185.95849609375, "completions/min_length": 90.6, "completions/min_terminated_length": 90.6, "epoch": 0.784, "grad_norm": 0.0012980833416804671, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 823464035.0, "reward": 1.1076786994934082, "reward_std": 0.08230673745274544, "rewards/accuracy_reward": 0.46953125, "rewards/brier_reward": 0.7831770896911621, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.45920702815055847, "rewards/smce": -0.07450821846723557, "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": 431.0, "completions/max_terminated_length": 431.0, "completions/mean_length": 190.05078125, "completions/mean_terminated_length": 190.05078125, "completions/min_length": 92.0, "completions/min_terminated_length": 92.0, "epoch": 0.8, "grad_norm": 0.0012545861536636949, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 840420715.0, "reward": 1.1329187154769897, "reward_std": 0.13519156575202942, "rewards/accuracy_reward": 0.49365234375, "rewards/brier_reward": 0.7884702205657959, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.506415045261383, "rewards/smce": -0.03257214445620775, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.002049180327868827, "eval_completions/max_length": 931.5, "eval_completions/max_terminated_length": 334.5, "eval_completions/mean_length": 196.33683013916016, "eval_completions/mean_terminated_length": 193.58879852294922, "eval_completions/min_length": 109.5, "eval_completions/min_terminated_length": 109.5, "eval_loss": 0.0, "eval_num_tokens": 840420715.0, "eval_reward": 1.0556547045707703, "eval_reward_std": 0.365094929933548, "eval_rewards/accuracy_reward": 0.376953125, "eval_rewards/brier_reward": 0.7935156226158142, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 0.998046875, "eval_rewards/mean_confidence_reward": 0.4902343899011612, "eval_rewards/smce": -0.11441406980156898, "eval_runtime": 22.1298, "eval_samples_per_second": 22.594, "eval_steps_per_second": 0.09, "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": 460.4, "completions/max_terminated_length": 460.4, "completions/mean_length": 191.8078125, "completions/mean_terminated_length": 191.8078125, "completions/min_length": 98.0, "completions/min_terminated_length": 98.0, "epoch": 0.816, "grad_norm": 0.0006870666402392089, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 857483995.0, "reward": 1.1305972576141357, "reward_std": 0.13415717780590058, "rewards/accuracy_reward": 0.48408203125, "rewards/brier_reward": 0.7903084397315979, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4990163028240204, "rewards/smce": -0.0261984433978796, "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": 662.6, "completions/max_terminated_length": 483.2, "completions/mean_length": 201.5015625, "completions/mean_terminated_length": 201.371142578125, "completions/min_length": 96.8, "completions/min_terminated_length": 96.8, "epoch": 0.832, "grad_norm": 0.0012498555006459355, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 874555723.0, "reward": 1.1258383750915528, "reward_std": 0.09739863574504852, "rewards/accuracy_reward": 0.48115234375, "rewards/brier_reward": 0.793649184703827, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.45032520294189454, "rewards/smce": -0.04605602435767651, "step": 260 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 772.8, "completions/max_terminated_length": 551.4, "completions/mean_length": 204.1986328125, "completions/mean_terminated_length": 204.06841125488282, "completions/min_length": 100.6, "completions/min_terminated_length": 100.6, "epoch": 0.848, "grad_norm": 0.0005774867604486644, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 891661085.0, "reward": 1.120431137084961, "reward_std": 0.09573130160570145, "rewards/accuracy_reward": 0.4623046875, "rewards/brier_reward": 0.7905663847923279, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44633203744888306, "rewards/smce": -0.023824118450284003, "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": 0.0001953125, "completions/max_length": 1015.2, "completions/max_terminated_length": 626.0, "completions/mean_length": 208.18095703125, "completions/mean_terminated_length": 207.9215057373047, "completions/min_length": 98.4, "completions/min_terminated_length": 98.4, "epoch": 0.864, "grad_norm": 0.0007092739688232541, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 908779674.0, "reward": 1.1397951364517211, "reward_std": 0.08750861436128617, "rewards/accuracy_reward": 0.50068359375, "rewards/brier_reward": 0.7907498121261597, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4849375069141388, "rewards/smce": -0.02329762177541852, "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": 0.00029296875, "completions/max_length": 940.4, "completions/max_terminated_length": 498.2, "completions/mean_length": 205.87783203125, "completions/mean_terminated_length": 205.48865661621093, "completions/min_length": 99.0, "completions/min_terminated_length": 99.0, "epoch": 0.88, "grad_norm": 0.0006288170116022229, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 926034935.0, "reward": 1.1090987920761108, "reward_std": 0.11937552094459533, "rewards/accuracy_reward": 0.44970703125, "rewards/brier_reward": 0.7989086747169495, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.49709376096725466, "rewards/smce": -0.06025214046239853, "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": 579.0, "completions/max_terminated_length": 579.0, "completions/mean_length": 208.03515625, "completions/mean_terminated_length": 208.03515625, "completions/min_length": 104.8, "completions/min_terminated_length": 104.8, "epoch": 0.896, "grad_norm": 0.0010682296706363559, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 943276063.0, "reward": 1.110878300666809, "reward_std": 0.15439056754112243, "rewards/accuracy_reward": 0.4568359375, "rewards/brier_reward": 0.792157506942749, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.510614275932312, "rewards/smce": -0.054475681856274605, "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": 822.8, "completions/max_terminated_length": 630.6, "completions/mean_length": 207.34013671875, "completions/mean_terminated_length": 207.2098815917969, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.912, "grad_norm": 0.0005791044677607715, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 960450522.0, "reward": 1.1124805688858033, "reward_std": 0.10075956732034683, "rewards/accuracy_reward": 0.4552734375, "rewards/brier_reward": 0.7934512853622436, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4501240313053131, "rewards/smce": -0.047138142958283426, "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": 9.765625e-05, "completions/max_length": 740.2, "completions/max_terminated_length": 519.0, "completions/mean_length": 201.753515625, "completions/mean_terminated_length": 201.6229217529297, "completions/min_length": 102.4, "completions/min_terminated_length": 102.4, "epoch": 0.928, "grad_norm": 0.0009222588851116598, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 977543294.0, "reward": 1.121965217590332, "reward_std": 0.08357778042554856, "rewards/accuracy_reward": 0.48759765625, "rewards/brier_reward": 0.7841106534004212, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.44317675828933717, "rewards/smce": -0.055362095683813096, "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": 0.000390625, "completions/max_length": 905.6, "completions/max_terminated_length": 643.6, "completions/mean_length": 201.8884765625, "completions/mean_terminated_length": 201.3676971435547, "completions/min_length": 91.2, "completions/min_terminated_length": 91.2, "epoch": 0.944, "grad_norm": 0.0006336426013149321, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 994586056.0, "reward": 1.1151194334030152, "reward_std": 0.10739713907241821, "rewards/accuracy_reward": 0.4759765625, "rewards/brier_reward": 0.7890912055969238, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4860449254512787, "rewards/smce": -0.0688786506652832, "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": 841.2, "completions/max_terminated_length": 659.8, "completions/mean_length": 205.77861328125, "completions/mean_terminated_length": 205.64851989746094, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.96, "grad_norm": 0.0012084617046639323, "learning_rate": 1e-06, "loss": -0.0004, "num_tokens": 1011633549.0, "reward": 1.1156575441360475, "reward_std": 0.13963824212551118, "rewards/accuracy_reward": 0.4591796875, "rewards/brier_reward": 0.7977803349494934, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.5077031373977661, "rewards/smce": -0.051096287369728086, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 495.0, "eval_completions/max_terminated_length": 495.0, "eval_completions/mean_length": 209.3858184814453, "eval_completions/mean_terminated_length": 209.3858184814453, "eval_completions/min_length": 114.5, "eval_completions/min_terminated_length": 114.5, "eval_loss": 0.0, "eval_num_tokens": 1011633549.0, "eval_reward": 1.0641183853149414, "eval_reward_std": 0.3622732162475586, "eval_rewards/accuracy_reward": 0.375, "eval_rewards/brier_reward": 0.8058867156505585, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4800390601158142, "eval_rewards/smce": -0.10530157387256622, "eval_runtime": 14.8357, "eval_samples_per_second": 33.702, "eval_steps_per_second": 0.135, "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": 912.4, "completions/max_terminated_length": 546.8, "completions/mean_length": 212.29677734375, "completions/mean_terminated_length": 211.9076354980469, "completions/min_length": 97.4, "completions/min_terminated_length": 97.4, "epoch": 0.976, "grad_norm": 0.0005940425908192992, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 1028668588.0, "reward": 1.1358384847640992, "reward_std": 0.07964012026786804, "rewards/accuracy_reward": 0.48720703125, "rewards/brier_reward": 0.7967997074127198, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.47436524033546446, "rewards/smce": -0.023880173824727534, "step": 305 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 964.4, "completions/max_terminated_length": 589.6, "completions/mean_length": 216.59365234375, "completions/mean_terminated_length": 216.3355255126953, "completions/min_length": 104.6, "completions/min_terminated_length": 104.6, "epoch": 0.992, "grad_norm": 0.0006660929648205638, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 1046014987.0, "reward": 1.1179794788360595, "reward_std": 0.1090321883559227, "rewards/accuracy_reward": 0.4705078125, "rewards/brier_reward": 0.7883034467697143, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4531269669532776, "rewards/smce": -0.04531567171216011, "step": 310 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 567.0, "completions/max_terminated_length": 567.0, "completions/mean_length": 215.76505279541016, "completions/mean_terminated_length": 215.76505279541016, "completions/min_length": 104.5, "completions/min_terminated_length": 104.5, "epoch": 0.9984, "num_tokens": 1052904757.0, "reward": 1.1193369626998901, "reward_std": 0.10750538855791092, "rewards/accuracy_reward": 0.467529296875, "rewards/brier_reward": 0.7913386523723602, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4925537258386612, "rewards/smce": -0.040389697067439556, "step": 312, "total_flos": 0.0, "train_loss": 0.003545210200872344, "train_runtime": 43249.5131, "train_samples_per_second": 0.462, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1052904757, "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 }