{ "best_global_step": null, "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": 0.0, "completions/max_length": 216.2, "completions/max_terminated_length": 216.2, "completions/mean_length": 92.0140625, "completions/mean_terminated_length": 92.0140625, "completions/min_length": 50.4, "completions/min_terminated_length": 50.4, "epoch": 0.016, "grad_norm": 0.25, "learning_rate": 3.1249999999999997e-07, "loss": -0.0017, "num_tokens": 196385.0, "reward": 0.5637666463851929, "reward_std": 0.09495568722486496, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.13846834152936935, "rewards/confidence_one_or_zero": 0.3265625, "rewards/format_reward": 0.9890625, "rewards/mean_confidence_reward": 0.913015627861023, "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": 0.0, "completions/max_length": 205.4, "completions/max_terminated_length": 205.4, "completions/mean_length": 90.140625, "completions/mean_terminated_length": 90.140625, "completions/min_length": 46.2, "completions/min_terminated_length": 46.2, "epoch": 0.032, "grad_norm": 0.1962890625, "learning_rate": 6.249999999999999e-07, "loss": -0.0022, "num_tokens": 391627.0, "reward": 0.5609725475311279, "reward_std": 0.10135613828897476, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.13600499778985978, "rewards/confidence_one_or_zero": 0.36875, "rewards/format_reward": 0.9859375, "rewards/mean_confidence_reward": 0.9128889083862305, "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": 0.0015625, "completions/max_length": 396.8, "completions/max_terminated_length": 222.8, "completions/mean_length": 94.0484375, "completions/mean_terminated_length": 92.58570404052735, "completions/min_length": 48.2, "completions/min_terminated_length": 48.2, "epoch": 0.048, "grad_norm": 0.2470703125, "learning_rate": 9.374999999999999e-07, "loss": -0.0024, "num_tokens": 589634.0, "reward": 0.5561538219451905, "reward_std": 0.09064869433641434, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.1279301553964615, "rewards/confidence_one_or_zero": 0.328125, "rewards/format_reward": 0.984375, "rewards/mean_confidence_reward": 0.920780622959137, "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": 0.0, "completions/max_length": 200.6, "completions/max_terminated_length": 200.6, "completions/mean_length": 92.79375, "completions/mean_terminated_length": 92.79375, "completions/min_length": 49.6, "completions/min_terminated_length": 49.6, "epoch": 0.064, "grad_norm": 0.30859375, "learning_rate": 1e-06, "loss": -0.0011, "num_tokens": 786758.0, "reward": 0.5728184819221497, "reward_std": 0.11178811639547348, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.16125953197479248, "rewards/confidence_one_or_zero": 0.2953125, "rewards/format_reward": 0.984375, "rewards/mean_confidence_reward": 0.893218731880188, "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": 0.0, "completions/max_length": 194.4, "completions/max_terminated_length": 194.4, "completions/mean_length": 92.2078125, "completions/mean_terminated_length": 92.2078125, "completions/min_length": 38.8, "completions/min_terminated_length": 38.8, "epoch": 0.08, "grad_norm": 0.27734375, "learning_rate": 1e-06, "loss": -0.0025, "num_tokens": 983563.0, "reward": 0.5665828108787536, "reward_std": 0.11537236422300338, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.15503820180892944, "rewards/confidence_one_or_zero": 0.3078125, "rewards/format_reward": 0.978125, "rewards/mean_confidence_reward": 0.898592185974121, "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": 0.0, "completions/max_length": 218.8, "completions/max_terminated_length": 218.8, "completions/mean_length": 93.2734375, "completions/mean_terminated_length": 93.2734375, "completions/min_length": 47.8, "completions/min_terminated_length": 47.8, "epoch": 0.096, "grad_norm": 0.287109375, "learning_rate": 1e-06, "loss": -0.003, "num_tokens": 1180666.0, "reward": 0.5757964849472046, "reward_std": 0.11430188566446305, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.17034062147140502, "rewards/confidence_one_or_zero": 0.2640625, "rewards/format_reward": 0.98125, "rewards/mean_confidence_reward": 0.8910624980926514, "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": 0.0015625, "completions/max_length": 362.8, "completions/max_terminated_length": 207.6, "completions/mean_length": 94.609375, "completions/mean_terminated_length": 93.16187286376953, "completions/min_length": 48.4, "completions/min_terminated_length": 48.4, "epoch": 0.112, "grad_norm": 0.255859375, "learning_rate": 1e-06, "loss": -0.0024, "num_tokens": 1379040.0, "reward": 0.5833104729652405, "reward_std": 0.11520791202783584, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.18536875247955323, "rewards/confidence_one_or_zero": 0.2140625, "rewards/format_reward": 0.98125, "rewards/mean_confidence_reward": 0.8816718816757202, "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": 0.0, "completions/max_length": 222.6, "completions/max_terminated_length": 222.6, "completions/mean_length": 94.7046875, "completions/mean_terminated_length": 94.7046875, "completions/min_length": 49.2, "completions/min_terminated_length": 49.2, "epoch": 0.128, "grad_norm": 0.208984375, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 1577123.0, "reward": 0.6028641343116761, "reward_std": 0.1150301843881607, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.21822625100612641, "rewards/confidence_one_or_zero": 0.15, "rewards/format_reward": 0.9875, "rewards/mean_confidence_reward": 0.8630921959877014, "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": 366.6, "completions/max_terminated_length": 204.2, "completions/mean_length": 96.8515625, "completions/mean_terminated_length": 95.40212860107422, "completions/min_length": 49.0, "completions/min_terminated_length": 49.0, "epoch": 0.144, "grad_norm": 0.2333984375, "learning_rate": 1e-06, "loss": -0.0004, "num_tokens": 1776684.0, "reward": 0.601399302482605, "reward_std": 0.11897070854902267, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.21529655754566193, "rewards/confidence_one_or_zero": 0.1546875, "rewards/format_reward": 0.9875, "rewards/mean_confidence_reward": 0.8657187342643737, "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": 0.0, "completions/max_length": 240.8, "completions/max_terminated_length": 240.8, "completions/mean_length": 97.65, "completions/mean_terminated_length": 97.65, "completions/min_length": 45.8, "completions/min_terminated_length": 45.8, "epoch": 0.16, "grad_norm": 0.244140625, "learning_rate": 1e-06, "loss": -0.0025, "num_tokens": 1976764.0, "reward": 0.623017156124115, "reward_std": 0.1165957733988762, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.2522823393344879, "rewards/confidence_one_or_zero": 0.1328125, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.8418124914169312, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 190.2, "eval_completions/max_terminated_length": 190.2, "eval_completions/mean_length": 96.7890625, "eval_completions/mean_terminated_length": 96.7890625, "eval_completions/min_length": 53.4, "eval_completions/min_terminated_length": 53.4, "eval_loss": 0.0, "eval_num_tokens": 1976764.0, "eval_reward": 0.6020447492599488, "eval_reward_std": 0.11144325025379657, "eval_rewards/accuracy_reward": 0.0, "eval_rewards/brier_reward": 0.21346249729394912, "eval_rewards/confidence_one_or_zero": 0.1375, "eval_rewards/format_reward": 0.990625, "eval_rewards/mean_confidence_reward": 0.8711843639612198, "eval_runtime": 89.5437, "eval_samples_per_second": 7.304, "eval_steps_per_second": 0.235, "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": 0.0, "completions/max_length": 226.8, "completions/max_terminated_length": 226.8, "completions/mean_length": 94.71875, "completions/mean_terminated_length": 94.71875, "completions/min_length": 51.2, "completions/min_terminated_length": 51.2, "epoch": 0.176, "grad_norm": 0.3046875, "learning_rate": 1e-06, "loss": -0.0009, "num_tokens": 2174944.0, "reward": 0.6206313490867614, "reward_std": 0.1254277005791664, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.25219823718070983, "rewards/confidence_one_or_zero": 0.14375, "rewards/format_reward": 0.9890625, "rewards/mean_confidence_reward": 0.8372890472412109, "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": 0.0, "completions/max_length": 233.4, "completions/max_terminated_length": 233.4, "completions/mean_length": 97.7734375, "completions/mean_terminated_length": 97.7734375, "completions/min_length": 50.0, "completions/min_terminated_length": 50.0, "epoch": 0.192, "grad_norm": 0.2421875, "learning_rate": 1e-06, "loss": -0.0025, "num_tokens": 2375319.0, "reward": 0.6321178793907165, "reward_std": 0.12163792401552201, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.2767339050769806, "rewards/confidence_one_or_zero": 0.0890625, "rewards/format_reward": 0.9875, "rewards/mean_confidence_reward": 0.8245624899864197, "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": 0.0, "completions/max_length": 230.6, "completions/max_terminated_length": 230.6, "completions/mean_length": 96.3421875, "completions/mean_terminated_length": 96.3421875, "completions/min_length": 51.0, "completions/min_terminated_length": 51.0, "epoch": 0.208, "grad_norm": 0.23046875, "learning_rate": 1e-06, "loss": -0.0016, "num_tokens": 2574546.0, "reward": 0.6542599558830261, "reward_std": 0.13078602403402328, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.31789315938949586, "rewards/confidence_one_or_zero": 0.09375, "rewards/format_reward": 0.990625, "rewards/mean_confidence_reward": 0.7956718564033508, "step": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 207.8, "completions/max_terminated_length": 207.8, "completions/mean_length": 96.878125, "completions/mean_terminated_length": 96.878125, "completions/min_length": 52.4, "completions/min_terminated_length": 52.4, "epoch": 0.224, "grad_norm": 0.255859375, "learning_rate": 1e-06, "loss": -0.0014, "num_tokens": 2774292.0, "reward": 0.6598511219024659, "reward_std": 0.12357937693595886, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.3243880569934845, "rewards/confidence_one_or_zero": 0.075, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.7928390502929688, "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": 0.0, "completions/max_length": 187.6, "completions/max_terminated_length": 187.6, "completions/mean_length": 96.1421875, "completions/mean_terminated_length": 96.1421875, "completions/min_length": 43.6, "completions/min_terminated_length": 43.6, "epoch": 0.24, "grad_norm": 0.267578125, "learning_rate": 1e-06, "loss": -0.0033, "num_tokens": 2973471.0, "reward": 0.6770576238632202, "reward_std": 0.13063670694828033, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.3588010907173157, "rewards/confidence_one_or_zero": 0.0546875, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.7654218554496766, "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": 219.2, "completions/max_terminated_length": 219.2, "completions/mean_length": 97.1765625, "completions/mean_terminated_length": 97.1765625, "completions/min_length": 48.6, "completions/min_terminated_length": 48.6, "epoch": 0.256, "grad_norm": 0.26171875, "learning_rate": 1e-06, "loss": -0.0002, "num_tokens": 3173248.0, "reward": 0.6767483830451966, "reward_std": 0.14175549149513245, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.3691201627254486, "rewards/confidence_one_or_zero": 0.05625, "rewards/format_reward": 0.984375, "rewards/mean_confidence_reward": 0.7536093473434449, "step": 80 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 221.2, "completions/max_terminated_length": 221.2, "completions/mean_length": 96.9, "completions/mean_terminated_length": 96.9, "completions/min_length": 47.4, "completions/min_terminated_length": 47.4, "epoch": 0.272, "grad_norm": 0.2333984375, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 3372992.0, "reward": 0.6882261753082275, "reward_std": 0.12886095941066741, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.38582578897476194, "rewards/confidence_one_or_zero": 0.05, "rewards/format_reward": 0.990625, "rewards/mean_confidence_reward": 0.747140622138977, "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": 291.0, "completions/max_terminated_length": 291.0, "completions/mean_length": 99.5171875, "completions/mean_terminated_length": 99.5171875, "completions/min_length": 52.4, "completions/min_terminated_length": 52.4, "epoch": 0.288, "grad_norm": 0.291015625, "learning_rate": 1e-06, "loss": -0.0004, "num_tokens": 3574507.0, "reward": 0.7160074472427368, "reward_std": 0.14012718051671982, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.4382634460926056, "rewards/confidence_one_or_zero": 0.04375, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.6999687433242798, "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": 0.0, "completions/max_length": 211.6, "completions/max_terminated_length": 211.6, "completions/mean_length": 99.64375, "completions/mean_terminated_length": 99.64375, "completions/min_length": 51.4, "completions/min_terminated_length": 51.4, "epoch": 0.304, "grad_norm": 0.28515625, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 3775967.0, "reward": 0.715096914768219, "reward_std": 0.13817149698734282, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.4411298394203186, "rewards/confidence_one_or_zero": 0.028125, "rewards/format_reward": 0.9890625, "rewards/mean_confidence_reward": 0.7031874895095825, "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": 0.0, "completions/max_length": 217.4, "completions/max_terminated_length": 217.4, "completions/mean_length": 101.6828125, "completions/mean_terminated_length": 101.6828125, "completions/min_length": 48.6, "completions/min_terminated_length": 48.6, "epoch": 0.32, "grad_norm": 0.283203125, "learning_rate": 1e-06, "loss": -0.0008, "num_tokens": 3978740.0, "reward": 0.7383638978004455, "reward_std": 0.13723695874214173, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.48141391277313234, "rewards/confidence_one_or_zero": 0.021875, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.6681562542915345, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 181.25, "eval_completions/max_terminated_length": 181.25, "eval_completions/mean_length": 98.940625, "eval_completions/mean_terminated_length": 98.940625, "eval_completions/min_length": 56.35, "eval_completions/min_terminated_length": 56.35, "eval_loss": 0.0, "eval_num_tokens": 3978740.0, "eval_reward": 0.7047684967517853, "eval_reward_std": 0.13906373605132102, "eval_rewards/accuracy_reward": 0.0, "eval_rewards/brier_reward": 0.4173479676246643, "eval_rewards/confidence_one_or_zero": 0.040625, "eval_rewards/format_reward": 0.9921875, "eval_rewards/mean_confidence_reward": 0.7250312507152558, "eval_runtime": 85.1017, "eval_samples_per_second": 7.685, "eval_steps_per_second": 0.247, "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": 0.0, "completions/max_length": 247.2, "completions/max_terminated_length": 247.2, "completions/mean_length": 95.7453125, "completions/mean_terminated_length": 95.7453125, "completions/min_length": 53.6, "completions/min_terminated_length": 53.6, "epoch": 0.336, "grad_norm": 0.30859375, "learning_rate": 1e-06, "loss": -0.0018, "num_tokens": 4177577.0, "reward": 0.7523154258728028, "reward_std": 0.14566269814968108, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.5124420523643494, "rewards/confidence_one_or_zero": 0.021875, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.6386093974113465, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 202.0, "completions/max_terminated_length": 202.0, "completions/mean_length": 96.8984375, "completions/mean_terminated_length": 96.8984375, "completions/min_length": 54.0, "completions/min_terminated_length": 54.0, "epoch": 0.352, "grad_norm": 0.26171875, "learning_rate": 1e-06, "loss": 0.0017, "num_tokens": 4377328.0, "reward": 0.7406684041023255, "reward_std": 0.15910515189170837, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.49539796710014344, "rewards/confidence_one_or_zero": 0.03125, "rewards/format_reward": 0.9859375, "rewards/mean_confidence_reward": 0.6531093716621399, "step": 110 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 224.4, "completions/max_terminated_length": 224.4, "completions/mean_length": 99.4375, "completions/mean_terminated_length": 99.4375, "completions/min_length": 48.0, "completions/min_terminated_length": 48.0, "epoch": 0.368, "grad_norm": 0.287109375, "learning_rate": 1e-06, "loss": -0.001, "num_tokens": 4578504.0, "reward": 0.7813051223754883, "reward_std": 0.1430785685777664, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.5672965407371521, "rewards/confidence_one_or_zero": 0.0203125, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.5921249866485596, "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": 0.0015625, "completions/max_length": 395.4, "completions/max_terminated_length": 231.8, "completions/mean_length": 99.5078125, "completions/mean_terminated_length": 98.05861206054688, "completions/min_length": 50.8, "completions/min_terminated_length": 50.8, "epoch": 0.384, "grad_norm": 0.291015625, "learning_rate": 1e-06, "loss": 0.0017, "num_tokens": 4779957.0, "reward": 0.7894837260246277, "reward_std": 0.15010789930820465, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.5867787599563599, "rewards/confidence_one_or_zero": 0.028125, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.5683437347412109, "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": 218.2, "completions/max_terminated_length": 218.2, "completions/mean_length": 95.478125, "completions/mean_terminated_length": 95.478125, "completions/min_length": 51.2, "completions/min_terminated_length": 51.2, "epoch": 0.4, "grad_norm": 0.306640625, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 4978679.0, "reward": 0.8271618366241456, "reward_std": 0.14937629997730256, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.6668226599693299, "rewards/confidence_one_or_zero": 0.0140625, "rewards/format_reward": 0.9875, "rewards/mean_confidence_reward": 0.4867968738079071, "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": 0.0, "completions/max_length": 232.0, "completions/max_terminated_length": 232.0, "completions/mean_length": 95.7359375, "completions/mean_terminated_length": 95.7359375, "completions/min_length": 50.6, "completions/min_terminated_length": 50.6, "epoch": 0.416, "grad_norm": 0.306640625, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 5177590.0, "reward": 0.8592821598052979, "reward_std": 0.12898499220609666, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.7216884374618531, "rewards/confidence_one_or_zero": 0.015625, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.43665624856948854, "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": 0.0, "completions/max_length": 178.8, "completions/max_terminated_length": 178.8, "completions/mean_length": 92.20625, "completions/mean_terminated_length": 92.20625, "completions/min_length": 50.6, "completions/min_terminated_length": 50.6, "epoch": 0.432, "grad_norm": 0.33203125, "learning_rate": 1e-06, "loss": 0.0018, "num_tokens": 5374058.0, "reward": 0.8699303984642028, "reward_std": 0.13423009365797042, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.7492349982261658, "rewards/confidence_one_or_zero": 0.00625, "rewards/format_reward": 0.990625, "rewards/mean_confidence_reward": 0.40295312404632566, "step": 135 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 178.2, "completions/max_terminated_length": 178.2, "completions/mean_length": 91.49375, "completions/mean_terminated_length": 91.49375, "completions/min_length": 43.4, "completions/min_terminated_length": 43.4, "epoch": 0.448, "grad_norm": 0.236328125, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 5570174.0, "reward": 0.8993057489395142, "reward_std": 0.11355664432048798, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.801735806465149, "rewards/confidence_one_or_zero": 0.0109375, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.3509531319141388, "step": 140 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 244.2, "completions/max_terminated_length": 244.2, "completions/mean_length": 90.7203125, "completions/mean_terminated_length": 90.7203125, "completions/min_length": 48.4, "completions/min_terminated_length": 48.4, "epoch": 0.464, "grad_norm": 0.314453125, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 5765971.0, "reward": 0.8959092974662781, "reward_std": 0.12455023974180221, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.801192831993103, "rewards/confidence_one_or_zero": 0.0265625, "rewards/format_reward": 0.990625, "rewards/mean_confidence_reward": 0.3403750121593475, "step": 145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 187.4, "completions/max_terminated_length": 187.4, "completions/mean_length": 88.0734375, "completions/mean_terminated_length": 88.0734375, "completions/min_length": 49.4, "completions/min_terminated_length": 49.4, "epoch": 0.48, "grad_norm": 0.2216796875, "learning_rate": 1e-06, "loss": 0.0027, "num_tokens": 5960082.0, "reward": 0.9339513182640076, "reward_std": 0.08665734976530075, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.8710270524024963, "rewards/confidence_one_or_zero": 0.0265625, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.26985938251018526, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 160.55, "eval_completions/max_terminated_length": 160.55, "eval_completions/mean_length": 91.8515625, "eval_completions/mean_terminated_length": 91.8515625, "eval_completions/min_length": 54.45, "eval_completions/min_terminated_length": 54.45, "eval_loss": 0.0, "eval_num_tokens": 5960082.0, "eval_reward": 0.9201600909233093, "eval_reward_std": 0.10864845588803292, "eval_rewards/accuracy_reward": 0.0, "eval_rewards/brier_reward": 0.8450070410966873, "eval_rewards/confidence_one_or_zero": 0.0109375, "eval_rewards/format_reward": 0.9953125, "eval_rewards/mean_confidence_reward": 0.30320313200354576, "eval_runtime": 75.8244, "eval_samples_per_second": 8.625, "eval_steps_per_second": 0.277, "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": 201.4, "completions/max_terminated_length": 201.4, "completions/mean_length": 86.7734375, "completions/mean_terminated_length": 86.7734375, "completions/min_length": 45.8, "completions/min_terminated_length": 45.8, "epoch": 0.496, "grad_norm": 0.15625, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 6153321.0, "reward": 0.9445138812065125, "reward_std": 0.08136400207877159, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.8921521902084351, "rewards/confidence_one_or_zero": 0.0140625, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.24190625548362732, "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": 209.6, "completions/max_terminated_length": 209.6, "completions/mean_length": 84.940625, "completions/mean_terminated_length": 84.940625, "completions/min_length": 49.2, "completions/min_terminated_length": 49.2, "epoch": 0.512, "grad_norm": 0.22265625, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 6345387.0, "reward": 0.9402939200401306, "reward_std": 0.09435945451259613, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.886837363243103, "rewards/confidence_one_or_zero": 0.0203125, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.23937343955039977, "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": 0.0015625, "completions/max_length": 388.2, "completions/max_terminated_length": 237.6, "completions/mean_length": 88.9734375, "completions/mean_terminated_length": 87.51290588378906, "completions/min_length": 48.6, "completions/min_terminated_length": 48.6, "epoch": 0.528, "grad_norm": 0.2099609375, "learning_rate": 1e-06, "loss": 0.0023, "num_tokens": 6539906.0, "reward": 0.9532355666160583, "reward_std": 0.0737876057624817, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9142831325531006, "rewards/confidence_one_or_zero": 0.0296875, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.20506250560283662, "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": 0.0015625, "completions/max_length": 341.6, "completions/max_terminated_length": 171.6, "completions/mean_length": 84.49375, "completions/mean_terminated_length": 83.02358551025391, "completions/min_length": 47.4, "completions/min_terminated_length": 47.4, "epoch": 0.544, "grad_norm": 0.109375, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 6731438.0, "reward": 0.955785083770752, "reward_std": 0.07366418913006782, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9193821907043457, "rewards/confidence_one_or_zero": 0.0328125, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.19231250286102294, "step": 170 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 212.0, "completions/max_terminated_length": 212.0, "completions/mean_length": 82.478125, "completions/mean_terminated_length": 82.478125, "completions/min_length": 48.0, "completions/min_terminated_length": 48.0, "epoch": 0.56, "grad_norm": 0.21484375, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 6921816.0, "reward": 0.9642942786216736, "reward_std": 0.057919394224882126, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9332756400108337, "rewards/confidence_one_or_zero": 0.028125, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.17956407070159913, "step": 175 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 177.6, "completions/max_terminated_length": 177.6, "completions/mean_length": 81.8671875, "completions/mean_terminated_length": 81.8671875, "completions/min_length": 46.6, "completions/min_terminated_length": 46.6, "epoch": 0.576, "grad_norm": 0.09912109375, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 7111555.0, "reward": 0.9653212189674377, "reward_std": 0.05407111048698425, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9337670564651489, "rewards/confidence_one_or_zero": 0.0234375, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.180796879529953, "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": 219.6, "completions/max_terminated_length": 219.6, "completions/mean_length": 83.190625, "completions/mean_terminated_length": 83.190625, "completions/min_length": 44.6, "completions/min_terminated_length": 44.6, "epoch": 0.592, "grad_norm": 0.07666015625, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 7302445.0, "reward": 0.9623199939727783, "reward_std": 0.06281839236617089, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9308895468711853, "rewards/confidence_one_or_zero": 0.01875, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.17901562750339509, "step": 185 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 175.8, "completions/max_terminated_length": 175.8, "completions/mean_length": 79.971875, "completions/mean_terminated_length": 79.971875, "completions/min_length": 41.4, "completions/min_terminated_length": 41.4, "epoch": 0.608, "grad_norm": 0.11767578125, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 7491331.0, "reward": 0.9639620184898376, "reward_std": 0.0702327698469162, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9372986197471619, "rewards/confidence_one_or_zero": 0.0203125, "rewards/format_reward": 0.990625, "rewards/mean_confidence_reward": 0.1677031248807907, "step": 190 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0015625, "completions/max_length": 344.2, "completions/max_terminated_length": 175.0, "completions/mean_length": 81.484375, "completions/mean_terminated_length": 80.00925140380859, "completions/min_length": 43.4, "completions/min_terminated_length": 43.4, "epoch": 0.624, "grad_norm": 0.296875, "learning_rate": 1e-06, "loss": 0.0018, "num_tokens": 7681129.0, "reward": 0.9704400300979614, "reward_std": 0.05550045445561409, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9486922025680542, "rewards/confidence_one_or_zero": 0.028125, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.15425000488758087, "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": 0.0015625, "completions/max_length": 336.0, "completions/max_terminated_length": 161.4, "completions/mean_length": 81.696875, "completions/mean_terminated_length": 80.21954956054688, "completions/min_length": 49.0, "completions/min_terminated_length": 49.0, "epoch": 0.64, "grad_norm": 0.1318359375, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 7871127.0, "reward": 0.9671568274497986, "reward_std": 0.061333130300045016, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9421258091926574, "rewards/confidence_one_or_zero": 0.0265625, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.158984375, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 152.6, "eval_completions/max_terminated_length": 152.6, "eval_completions/mean_length": 86.640625, "eval_completions/mean_terminated_length": 86.640625, "eval_completions/min_length": 52.75, "eval_completions/min_terminated_length": 52.75, "eval_loss": 0.0, "eval_num_tokens": 7871127.0, "eval_reward": 0.9708386063575745, "eval_reward_std": 0.05269177481532097, "eval_rewards/accuracy_reward": 0.0, "eval_rewards/brier_reward": 0.9448018282651901, "eval_rewards/confidence_one_or_zero": 0.0171875, "eval_rewards/format_reward": 0.996875, "eval_rewards/mean_confidence_reward": 0.16914578452706336, "eval_runtime": 72.3018, "eval_samples_per_second": 9.045, "eval_steps_per_second": 0.29, "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": 157.2, "completions/max_terminated_length": 157.2, "completions/mean_length": 78.59375, "completions/mean_terminated_length": 78.59375, "completions/min_length": 44.0, "completions/min_terminated_length": 44.0, "epoch": 0.656, "grad_norm": 0.0498046875, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 8059035.0, "reward": 0.976018238067627, "reward_std": 0.0433214133605361, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9551610946655273, "rewards/confidence_one_or_zero": 0.0375, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.14848437607288362, "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": 0.0, "completions/max_length": 185.2, "completions/max_terminated_length": 185.2, "completions/mean_length": 79.2890625, "completions/mean_terminated_length": 79.2890625, "completions/min_length": 45.6, "completions/min_terminated_length": 45.6, "epoch": 0.672, "grad_norm": 0.1025390625, "learning_rate": 1e-06, "loss": 0.0017, "num_tokens": 8247404.0, "reward": 0.9767295479774475, "reward_std": 0.041362495720386507, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9581462502479553, "rewards/confidence_one_or_zero": 0.0359375, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.14453125, "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": 175.4, "completions/max_terminated_length": 175.4, "completions/mean_length": 79.659375, "completions/mean_terminated_length": 79.659375, "completions/min_length": 38.0, "completions/min_terminated_length": 38.0, "epoch": 0.688, "grad_norm": 0.228515625, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 8435754.0, "reward": 0.971716296672821, "reward_std": 0.050771788507699964, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9528071999549865, "rewards/confidence_one_or_zero": 0.0234375, "rewards/format_reward": 0.990625, "rewards/mean_confidence_reward": 0.1443750038743019, "step": 215 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 166.2, "completions/max_terminated_length": 166.2, "completions/mean_length": 78.75, "completions/mean_terminated_length": 78.75, "completions/min_length": 45.8, "completions/min_terminated_length": 45.8, "epoch": 0.704, "grad_norm": 0.0888671875, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 8623874.0, "reward": 0.9842060446739197, "reward_std": 0.02394692096859217, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9684117436408997, "rewards/confidence_one_or_zero": 0.025, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.13839063048362732, "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": 0.0, "completions/max_length": 178.2, "completions/max_terminated_length": 178.2, "completions/mean_length": 78.165625, "completions/mean_terminated_length": 78.165625, "completions/min_length": 45.6, "completions/min_terminated_length": 45.6, "epoch": 0.72, "grad_norm": 0.154296875, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 8811404.0, "reward": 0.9803202271461486, "reward_std": 0.03289288468658924, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.963765025138855, "rewards/confidence_one_or_zero": 0.0453125, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.1364375039935112, "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": 0.0, "completions/max_length": 183.4, "completions/max_terminated_length": 183.4, "completions/mean_length": 79.3640625, "completions/mean_terminated_length": 79.3640625, "completions/min_length": 46.4, "completions/min_terminated_length": 46.4, "epoch": 0.736, "grad_norm": 0.17578125, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 8999829.0, "reward": 0.9792708635330201, "reward_std": 0.037150106951594354, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9616663455963135, "rewards/confidence_one_or_zero": 0.028125, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.13943437337875367, "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": 146.2, "completions/max_terminated_length": 146.2, "completions/mean_length": 76.91875, "completions/mean_terminated_length": 76.91875, "completions/min_length": 46.0, "completions/min_terminated_length": 46.0, "epoch": 0.752, "grad_norm": 0.05029296875, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 9186593.0, "reward": 0.9837331652641297, "reward_std": 0.028783718310296535, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9705910563468934, "rewards/confidence_one_or_zero": 0.0265625, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.12692468613386154, "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": 0.0015625, "completions/max_length": 342.0, "completions/max_terminated_length": 183.0, "completions/mean_length": 79.4609375, "completions/mean_terminated_length": 77.98030242919921, "completions/min_length": 45.4, "completions/min_terminated_length": 45.4, "epoch": 0.768, "grad_norm": 0.068359375, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 9374880.0, "reward": 0.9828169703483581, "reward_std": 0.028539419639855623, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9687586188316345, "rewards/confidence_one_or_zero": 0.0234375, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.12948437929153442, "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": 177.0, "completions/max_terminated_length": 177.0, "completions/mean_length": 79.0390625, "completions/mean_terminated_length": 79.0390625, "completions/min_length": 49.0, "completions/min_terminated_length": 49.0, "epoch": 0.784, "grad_norm": 0.08447265625, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 9563289.0, "reward": 0.9848930597305298, "reward_std": 0.02363475486636162, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9697857856750488, "rewards/confidence_one_or_zero": 0.0296875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.1321406215429306, "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": 297.2, "completions/max_terminated_length": 297.2, "completions/mean_length": 77.6859375, "completions/mean_terminated_length": 77.6859375, "completions/min_length": 46.0, "completions/min_terminated_length": 46.0, "epoch": 0.8, "grad_norm": 0.091796875, "learning_rate": 1e-06, "loss": 0.0024, "num_tokens": 9750760.0, "reward": 0.9798139333724976, "reward_std": 0.03771413471549749, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9643150091171264, "rewards/confidence_one_or_zero": 0.03125, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.133062507212162, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 155.4, "eval_completions/max_terminated_length": 155.4, "eval_completions/mean_length": 83.25625, "eval_completions/mean_terminated_length": 83.25625, "eval_completions/min_length": 51.5, "eval_completions/min_terminated_length": 51.5, "eval_loss": 0.0, "eval_num_tokens": 9750760.0, "eval_reward": 0.9758151024580002, "eval_reward_std": 0.04481957714888267, "eval_rewards/accuracy_reward": 0.0, "eval_rewards/brier_reward": 0.9547548532485962, "eval_rewards/confidence_one_or_zero": 0.0125, "eval_rewards/format_reward": 0.996875, "eval_rewards/mean_confidence_reward": 0.15020312890410423, "eval_runtime": 73.5394, "eval_samples_per_second": 8.893, "eval_steps_per_second": 0.286, "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": 158.2, "completions/max_terminated_length": 158.2, "completions/mean_length": 75.653125, "completions/mean_terminated_length": 75.653125, "completions/min_length": 38.0, "completions/min_terminated_length": 38.0, "epoch": 0.816, "grad_norm": 0.11474609375, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 9936906.0, "reward": 0.9771119117736816, "reward_std": 0.04261733442544937, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9604734539985657, "rewards/confidence_one_or_zero": 0.03125, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.13475000262260436, "step": 255 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0015625, "completions/max_length": 349.6, "completions/max_terminated_length": 169.0, "completions/mean_length": 77.9703125, "completions/mean_terminated_length": 76.489111328125, "completions/min_length": 44.4, "completions/min_terminated_length": 44.4, "epoch": 0.832, "grad_norm": 0.0556640625, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 10124607.0, "reward": 0.9843807220458984, "reward_std": 0.026195484772324563, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.971886134147644, "rewards/confidence_one_or_zero": 0.0265625, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.12435937821865081, "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": 0.0, "completions/max_length": 170.6, "completions/max_terminated_length": 170.6, "completions/mean_length": 77.63125, "completions/mean_terminated_length": 77.63125, "completions/min_length": 45.8, "completions/min_terminated_length": 45.8, "epoch": 0.848, "grad_norm": 0.0830078125, "learning_rate": 1e-06, "loss": 0.0019, "num_tokens": 10312059.0, "reward": 0.9819784879684448, "reward_std": 0.03134257197380066, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9686441779136657, "rewards/confidence_one_or_zero": 0.03125, "rewards/format_reward": 0.9953125, "rewards/mean_confidence_reward": 0.12544219046831132, "step": 265 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 150.4, "completions/max_terminated_length": 150.4, "completions/mean_length": 75.5765625, "completions/mean_terminated_length": 75.5765625, "completions/min_length": 44.0, "completions/min_terminated_length": 44.0, "epoch": 0.864, "grad_norm": 0.0986328125, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 10498092.0, "reward": 0.984512984752655, "reward_std": 0.023365317098796367, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9690256357192993, "rewards/confidence_one_or_zero": 0.0296875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.13040625005960466, "step": 270 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 175.4, "completions/max_terminated_length": 175.4, "completions/mean_length": 76.9015625, "completions/mean_terminated_length": 76.9015625, "completions/min_length": 43.2, "completions/min_terminated_length": 43.2, "epoch": 0.88, "grad_norm": 0.1181640625, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 10684973.0, "reward": 0.9863678097724915, "reward_std": 0.0205452554859221, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9742978334426879, "rewards/confidence_one_or_zero": 0.0390625, "rewards/format_reward": 0.9984375, "rewards/mean_confidence_reward": 0.12199999988079072, "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": 0.0015625, "completions/max_length": 335.0, "completions/max_terminated_length": 164.4, "completions/mean_length": 78.6171875, "completions/mean_terminated_length": 77.13751220703125, "completions/min_length": 45.4, "completions/min_terminated_length": 45.4, "epoch": 0.896, "grad_norm": 0.045654296875, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 10872912.0, "reward": 0.9855192422866821, "reward_std": 0.023602598533034324, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9741631031036377, "rewards/confidence_one_or_zero": 0.0390625, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.12093281596899033, "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": 0.0, "completions/max_length": 192.4, "completions/max_terminated_length": 192.4, "completions/mean_length": 76.009375, "completions/mean_terminated_length": 76.009375, "completions/min_length": 46.0, "completions/min_terminated_length": 46.0, "epoch": 0.912, "grad_norm": 0.01263427734375, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 11059174.0, "reward": 0.9828890323638916, "reward_std": 0.03203322989866138, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9735901713371277, "rewards/confidence_one_or_zero": 0.034375, "rewards/format_reward": 0.9921875, "rewards/mean_confidence_reward": 0.11648437529802322, "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": 161.6, "completions/max_terminated_length": 161.6, "completions/mean_length": 76.6296875, "completions/mean_terminated_length": 76.6296875, "completions/min_length": 46.4, "completions/min_terminated_length": 46.4, "epoch": 0.928, "grad_norm": 0.1318359375, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 11245817.0, "reward": 0.9842098593711853, "reward_std": 0.02493043690919876, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9715443849563599, "rewards/confidence_one_or_zero": 0.03125, "rewards/format_reward": 0.996875, "rewards/mean_confidence_reward": 0.12405781298875809, "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": 0.0015625, "completions/max_length": 332.2, "completions/max_terminated_length": 183.0, "completions/mean_length": 76.5609375, "completions/mean_terminated_length": 75.07853088378906, "completions/min_length": 43.2, "completions/min_terminated_length": 43.2, "epoch": 0.944, "grad_norm": 0.203125, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 11432408.0, "reward": 0.9761037707328797, "reward_std": 0.04192298874258995, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9584571957588196, "rewards/confidence_one_or_zero": 0.028125, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.13474999964237214, "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": 0.0015625, "completions/max_length": 340.4, "completions/max_terminated_length": 177.0, "completions/mean_length": 77.4609375, "completions/mean_terminated_length": 75.98180389404297, "completions/min_length": 45.0, "completions/min_terminated_length": 45.0, "epoch": 0.96, "grad_norm": 0.1650390625, "learning_rate": 1e-06, "loss": 0.0017, "num_tokens": 11619671.0, "reward": 0.9803122878074646, "reward_std": 0.03791804723441601, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9668742299079895, "rewards/confidence_one_or_zero": 0.0375, "rewards/format_reward": 0.99375, "rewards/mean_confidence_reward": 0.12432812452316284, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 144.3, "eval_completions/max_terminated_length": 144.3, "eval_completions/mean_length": 80.825, "eval_completions/mean_terminated_length": 80.825, "eval_completions/min_length": 51.3, "eval_completions/min_terminated_length": 51.3, "eval_loss": 0.0, "eval_num_tokens": 11619671.0, "eval_reward": 0.9829227596521377, "eval_reward_std": 0.03057727730483748, "eval_rewards/accuracy_reward": 0.0, "eval_rewards/brier_reward": 0.9689701855182647, "eval_rewards/confidence_one_or_zero": 0.0265625, "eval_rewards/format_reward": 0.996875, "eval_rewards/mean_confidence_reward": 0.12760937735438346, "eval_runtime": 68.6291, "eval_samples_per_second": 9.529, "eval_steps_per_second": 0.306, "step": 300 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 169.0, "completions/max_terminated_length": 169.0, "completions/mean_length": 75.484375, "completions/mean_terminated_length": 75.484375, "completions/min_length": 45.2, "completions/min_terminated_length": 45.2, "epoch": 0.976, "grad_norm": 0.0238037109375, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 11805685.0, "reward": 0.9886146664619446, "reward_std": 0.016557221487164497, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9772288918495178, "rewards/confidence_one_or_zero": 0.046875, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.11732812225818634, "step": 305 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0, "completions/max_length": 162.6, "completions/max_terminated_length": 162.6, "completions/mean_length": 77.559375, "completions/mean_terminated_length": 77.559375, "completions/min_length": 42.2, "completions/min_terminated_length": 42.2, "epoch": 0.992, "grad_norm": 0.045166015625, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 11993139.0, "reward": 0.9865232825279235, "reward_std": 0.02098242472857237, "rewards/accuracy_reward": 0.0, "rewards/brier_reward": 0.9746087789535522, "rewards/confidence_one_or_zero": 0.0296875, "rewards/format_reward": 0.9984375, "rewards/mean_confidence_reward": 0.12168750315904617, "step": 310 }, { "epoch": 0.9984, "step": 312, "total_flos": 0.0, "train_loss": 0.0, "train_runtime": 1.7091, "train_samples_per_second": 2925.582, "train_steps_per_second": 182.556 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 12067150, "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": 2, "trial_name": null, "trial_params": null }