{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.015356265356265357, "eval_steps": 500, "global_step": 100, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.00015356265356265356, "frac_reward_zero_std": 1.0, "grad_norm": 0.0, "kl": 0.0, "learning_rate": 0.0, "loss": 0.0, "num_tokens": 2528.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 1 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0003071253071253071, "frac_reward_zero_std": 0.0, "grad_norm": 0.22934769093990326, "kl": 0.0, "learning_rate": 5.0000000000000004e-08, "loss": 0.0, "num_tokens": 5056.0, "reward": 0.75, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.4629100561141968, "step": 2 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0004606879606879607, "frac_reward_zero_std": 1.0, "grad_norm": 2.4884398953872733e-05, "kl": 0.0011284813517704606, "learning_rate": 1.0000000000000001e-07, "loss": 0.0, "num_tokens": 7584.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 3 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0006142506142506142, "frac_reward_zero_std": 1.0, "grad_norm": 5.449208038044162e-05, "kl": 0.0020245485939085484, "learning_rate": 1.5000000000000002e-07, "loss": 0.0, "num_tokens": 10112.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 4 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0007678132678132678, "frac_reward_zero_std": 0.0, "grad_norm": 0.28413450717926025, "kl": 0.0017015498597174883, "learning_rate": 2.0000000000000002e-07, "loss": 0.0, "num_tokens": 12640.0, "reward": 0.75, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.4629100561141968, "step": 5 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0009213759213759214, "frac_reward_zero_std": 1.0, "grad_norm": 4.5294800656847656e-05, "kl": 0.001824827864766121, "learning_rate": 2.5000000000000004e-07, "loss": 0.0, "num_tokens": 15168.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 6 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.001074938574938575, "frac_reward_zero_std": 0.0, "grad_norm": 0.22941677272319794, "kl": 0.001545972190797329, "learning_rate": 3.0000000000000004e-07, "loss": 0.0, "num_tokens": 17696.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 7 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0012285012285012285, "frac_reward_zero_std": 0.0, "grad_norm": 0.24631166458129883, "kl": 0.0017354765441268682, "learning_rate": 3.5000000000000004e-07, "loss": 0.0, "num_tokens": 20224.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 8 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.001382063882063882, "frac_reward_zero_std": 1.0, "grad_norm": 4.0858907595975325e-05, "kl": 0.0018943097675219178, "learning_rate": 4.0000000000000003e-07, "loss": 0.0, "num_tokens": 22752.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 9 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0015356265356265355, "frac_reward_zero_std": 1.0, "grad_norm": 0.00010698483674786985, "kl": 0.0019578838255256414, "learning_rate": 4.5000000000000003e-07, "loss": 0.0, "num_tokens": 25280.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 10 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0016891891891891893, "frac_reward_zero_std": 1.0, "grad_norm": 3.508454028633423e-05, "kl": 0.001726516056805849, "learning_rate": 5.000000000000001e-07, "loss": 0.0, "num_tokens": 27808.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 11 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0018427518427518428, "frac_reward_zero_std": 0.0, "grad_norm": 0.26142945885658264, "kl": 0.001936572720296681, "learning_rate": 5.5e-07, "loss": 0.0, "num_tokens": 30336.0, "reward": 0.5, "reward_std": 0.5345224738121033, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5345224738121033, "step": 12 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0019963144963144963, "frac_reward_zero_std": 1.0, "grad_norm": 3.853518865071237e-05, "kl": 0.001674045342952013, "learning_rate": 6.000000000000001e-07, "loss": 0.0, "num_tokens": 32864.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 13 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.00214987714987715, "frac_reward_zero_std": 1.0, "grad_norm": 5.5004267778713256e-05, "kl": 0.0016993756871670485, "learning_rate": 6.5e-07, "loss": 0.0, "num_tokens": 35392.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 14 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0023034398034398034, "frac_reward_zero_std": 1.0, "grad_norm": 0.000111581408418715, "kl": 0.0021551968529820442, "learning_rate": 7.000000000000001e-07, "loss": 0.0, "num_tokens": 37920.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 15 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.002457002457002457, "frac_reward_zero_std": 0.0, "grad_norm": 0.2612421214580536, "kl": 0.0017377862241119146, "learning_rate": 7.5e-07, "loss": 0.0, "num_tokens": 40448.0, "reward": 0.25, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.4629100561141968, "step": 16 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0026105651105651105, "frac_reward_zero_std": 1.0, "grad_norm": 4.1293584217783064e-05, "kl": 0.00191373226698488, "learning_rate": 8.000000000000001e-07, "loss": 0.0, "num_tokens": 42976.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 17 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.002764127764127764, "frac_reward_zero_std": 1.0, "grad_norm": 0.00013062957441434264, "kl": 0.0017311959527432919, "learning_rate": 8.500000000000001e-07, "loss": 0.0, "num_tokens": 45504.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 18 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0029176904176904175, "frac_reward_zero_std": 1.0, "grad_norm": 4.6584205847466365e-05, "kl": 0.001798063050955534, "learning_rate": 9.000000000000001e-07, "loss": 0.0, "num_tokens": 48032.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 19 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.003071253071253071, "frac_reward_zero_std": 1.0, "grad_norm": 4.3272721086395904e-05, "kl": 0.0018202185165137053, "learning_rate": 9.500000000000001e-07, "loss": 0.0, "num_tokens": 50560.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 20 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0032248157248157246, "frac_reward_zero_std": 1.0, "grad_norm": 7.859696779632941e-05, "kl": 0.0018265624530613422, "learning_rate": 1.0000000000000002e-06, "loss": 0.0, "num_tokens": 53088.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 21 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0033783783783783786, "frac_reward_zero_std": 0.0, "grad_norm": 0.26943519711494446, "kl": 0.0018876121612265706, "learning_rate": 1.0500000000000001e-06, "loss": 0.0, "num_tokens": 55616.0, "reward": 0.25, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.4629100561141968, "step": 22 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.003531941031941032, "frac_reward_zero_std": 0.0, "grad_norm": 0.203104630112648, "kl": 0.0015638768672943115, "learning_rate": 1.1e-06, "loss": 0.0, "num_tokens": 58144.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 23 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0036855036855036856, "frac_reward_zero_std": 1.0, "grad_norm": 4.421776611707173e-05, "kl": 0.0017870652955025434, "learning_rate": 1.1500000000000002e-06, "loss": 0.0, "num_tokens": 60672.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 24 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.003839066339066339, "frac_reward_zero_std": 0.0, "grad_norm": 0.26979029178619385, "kl": 0.001943537499755621, "learning_rate": 1.2000000000000002e-06, "loss": 0.0, "num_tokens": 63200.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 25 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.003992628992628993, "frac_reward_zero_std": 0.0, "grad_norm": 0.27013102173805237, "kl": 0.001535331830382347, "learning_rate": 1.25e-06, "loss": 0.0, "num_tokens": 65728.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 26 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.004146191646191646, "frac_reward_zero_std": 1.0, "grad_norm": 3.67719039786607e-05, "kl": 0.0017323686042800546, "learning_rate": 1.3e-06, "loss": 0.0, "num_tokens": 68256.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 27 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0042997542997543, "frac_reward_zero_std": 0.0, "grad_norm": 0.23535043001174927, "kl": 0.0020425482653081417, "learning_rate": 1.3500000000000002e-06, "loss": 0.0, "num_tokens": 70784.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 28 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.004453316953316954, "frac_reward_zero_std": 1.0, "grad_norm": 3.760838080779649e-05, "kl": 0.0016436483711004257, "learning_rate": 1.4000000000000001e-06, "loss": 0.0, "num_tokens": 73312.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 29 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.004606879606879607, "frac_reward_zero_std": 0.0, "grad_norm": 0.25959762930870056, "kl": 0.0017603226006031036, "learning_rate": 1.45e-06, "loss": 0.0, "num_tokens": 75840.0, "reward": 0.375, "reward_std": 0.5175491571426392, "rewards/compute_reward/mean": 0.375, "rewards/compute_reward/std": 0.5175492167472839, "step": 30 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.004760442260442261, "frac_reward_zero_std": 0.0, "grad_norm": 0.23616784811019897, "kl": 0.0018221420468762517, "learning_rate": 1.5e-06, "loss": 0.0, "num_tokens": 78368.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 31 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.004914004914004914, "frac_reward_zero_std": 1.0, "grad_norm": 4.308336792746559e-05, "kl": 0.001862606150098145, "learning_rate": 1.5500000000000002e-06, "loss": 0.0, "num_tokens": 80896.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 32 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005067567567567568, "frac_reward_zero_std": 1.0, "grad_norm": 4.589068339555524e-05, "kl": 0.001723211957141757, "learning_rate": 1.6000000000000001e-06, "loss": 0.0, "num_tokens": 83424.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 33 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005221130221130221, "frac_reward_zero_std": 0.0, "grad_norm": 0.30930790305137634, "kl": 0.001744288601912558, "learning_rate": 1.6500000000000003e-06, "loss": 0.0, "num_tokens": 85952.0, "reward": 0.25, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.4629100561141968, "step": 34 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005374692874692875, "frac_reward_zero_std": 1.0, "grad_norm": 5.498694008565508e-05, "kl": 0.001765427179634571, "learning_rate": 1.7000000000000002e-06, "loss": 0.0, "num_tokens": 88480.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 35 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005528255528255528, "frac_reward_zero_std": 0.0, "grad_norm": 0.45219364762306213, "kl": 0.0013709957711398602, "learning_rate": 1.75e-06, "loss": 0.0, "num_tokens": 91008.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 36 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005681818181818182, "frac_reward_zero_std": 1.0, "grad_norm": 5.568007691181265e-05, "kl": 0.0015674353344365954, "learning_rate": 1.8000000000000001e-06, "loss": 0.0, "num_tokens": 93536.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 37 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005835380835380835, "frac_reward_zero_std": 1.0, "grad_norm": 0.00022698717657476664, "kl": 0.002098341705277562, "learning_rate": 1.85e-06, "loss": 0.0, "num_tokens": 96064.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 38 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.005988943488943489, "frac_reward_zero_std": 1.0, "grad_norm": 4.615697616827674e-05, "kl": 0.0018569512758404016, "learning_rate": 1.9000000000000002e-06, "loss": 0.0, "num_tokens": 98592.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 39 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.006142506142506142, "frac_reward_zero_std": 0.0, "grad_norm": 0.3124924898147583, "kl": 0.002205869182944298, "learning_rate": 1.9500000000000004e-06, "loss": 0.0, "num_tokens": 101120.0, "reward": 0.75, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.75, "rewards/compute_reward/std": 0.4629100561141968, "step": 40 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.006296068796068796, "frac_reward_zero_std": 1.0, "grad_norm": 4.875810554949567e-05, "kl": 0.001907296129502356, "learning_rate": 2.0000000000000003e-06, "loss": 0.0, "num_tokens": 103648.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 41 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.006449631449631449, "frac_reward_zero_std": 1.0, "grad_norm": 4.659927071770653e-05, "kl": 0.0018478649435564876, "learning_rate": 2.05e-06, "loss": 0.0, "num_tokens": 106176.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 42 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.006603194103194103, "frac_reward_zero_std": 1.0, "grad_norm": 8.290458208648488e-05, "kl": 0.00224427436478436, "learning_rate": 2.1000000000000002e-06, "loss": 0.0, "num_tokens": 108704.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 43 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.006756756756756757, "frac_reward_zero_std": 1.0, "grad_norm": 5.3578805818688124e-05, "kl": 0.0018698148196563125, "learning_rate": 2.15e-06, "loss": 0.0, "num_tokens": 111232.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 44 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.00691031941031941, "frac_reward_zero_std": 1.0, "grad_norm": 4.0759008697932586e-05, "kl": 0.0018630733247846365, "learning_rate": 2.2e-06, "loss": 0.0, "num_tokens": 113760.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 45 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007063882063882064, "frac_reward_zero_std": 1.0, "grad_norm": 4.1463466914137825e-05, "kl": 0.0014028309378772974, "learning_rate": 2.25e-06, "loss": 0.0, "num_tokens": 116288.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 46 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007217444717444717, "frac_reward_zero_std": 1.0, "grad_norm": 5.521080311154947e-05, "kl": 0.0018875582609325647, "learning_rate": 2.3000000000000004e-06, "loss": 0.0, "num_tokens": 118816.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 47 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007371007371007371, "frac_reward_zero_std": 0.0, "grad_norm": 0.27408236265182495, "kl": 0.002105266787111759, "learning_rate": 2.35e-06, "loss": 0.0, "num_tokens": 121344.0, "reward": 0.25, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.4629100561141968, "step": 48 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007524570024570024, "frac_reward_zero_std": 1.0, "grad_norm": 0.00027501373551785946, "kl": 0.0023503885604441166, "learning_rate": 2.4000000000000003e-06, "loss": 0.0, "num_tokens": 123872.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 49 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007678132678132678, "frac_reward_zero_std": 1.0, "grad_norm": 4.1167542804032564e-05, "kl": 0.0015902362065389752, "learning_rate": 2.4500000000000003e-06, "loss": 0.0, "num_tokens": 126400.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 50 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007831695331695332, "frac_reward_zero_std": 1.0, "grad_norm": 6.368236790876836e-05, "kl": 0.0016205956926569343, "learning_rate": 2.5e-06, "loss": 0.0, "num_tokens": 128928.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 51 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.007985257985257985, "frac_reward_zero_std": 0.0, "grad_norm": 0.2795107364654541, "kl": 0.0019439270254224539, "learning_rate": 2.55e-06, "loss": 0.0, "num_tokens": 131456.0, "reward": 0.5, "reward_std": 0.5345224738121033, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5345224738121033, "step": 52 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.008138820638820638, "frac_reward_zero_std": 0.0, "grad_norm": 0.2855827510356903, "kl": 0.0019104867242276669, "learning_rate": 2.6e-06, "loss": 0.0, "num_tokens": 133984.0, "reward": 0.25, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.4629100561141968, "step": 53 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.008292383292383292, "frac_reward_zero_std": 1.0, "grad_norm": 6.060176747268997e-05, "kl": 0.0017747373785823584, "learning_rate": 2.6500000000000005e-06, "loss": 0.0, "num_tokens": 136512.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 54 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.008445945945945946, "frac_reward_zero_std": 0.0, "grad_norm": 0.28289034962654114, "kl": 0.0022900549229234457, "learning_rate": 2.7000000000000004e-06, "loss": 0.0, "num_tokens": 139040.0, "reward": 0.625, "reward_std": 0.5175491571426392, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5175492167472839, "step": 55 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.0085995085995086, "frac_reward_zero_std": 1.0, "grad_norm": 4.690838613896631e-05, "kl": 0.001854082802310586, "learning_rate": 2.7500000000000004e-06, "loss": 0.0, "num_tokens": 141568.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 56 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.008753071253071253, "frac_reward_zero_std": 1.0, "grad_norm": 4.404248102218844e-05, "kl": 0.0015642520738765597, "learning_rate": 2.8000000000000003e-06, "loss": 0.0, "num_tokens": 144096.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 57 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.008906633906633907, "frac_reward_zero_std": 0.0, "grad_norm": 0.2607027590274811, "kl": 0.001625827164389193, "learning_rate": 2.85e-06, "loss": 0.0, "num_tokens": 146624.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 58 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.00906019656019656, "frac_reward_zero_std": 1.0, "grad_norm": 4.736987830256112e-05, "kl": 0.0018028158228844404, "learning_rate": 2.9e-06, "loss": 0.0, "num_tokens": 149152.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 59 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.009213759213759214, "frac_reward_zero_std": 0.0, "grad_norm": 0.2748159170150757, "kl": 0.0017374631715938449, "learning_rate": 2.95e-06, "loss": 0.0, "num_tokens": 151680.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 60 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.009367321867321867, "frac_reward_zero_std": 1.0, "grad_norm": 4.3023454054491594e-05, "kl": 0.00184379902202636, "learning_rate": 3e-06, "loss": 0.0, "num_tokens": 154208.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 61 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.009520884520884522, "frac_reward_zero_std": 1.0, "grad_norm": 3.636611290858127e-05, "kl": 0.001865661353804171, "learning_rate": 3.05e-06, "loss": 0.0, "num_tokens": 156736.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 62 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.009674447174447175, "frac_reward_zero_std": 1.0, "grad_norm": 4.183187411399558e-05, "kl": 0.0015882912557572126, "learning_rate": 3.1000000000000004e-06, "loss": 0.0, "num_tokens": 159264.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 63 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.009828009828009828, "frac_reward_zero_std": 1.0, "grad_norm": 3.482505780993961e-05, "kl": 0.0016557248309254646, "learning_rate": 3.1500000000000003e-06, "loss": 0.0, "num_tokens": 161792.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 64 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.00998157248157248, "frac_reward_zero_std": 1.0, "grad_norm": 4.539267683867365e-05, "kl": 0.0019870202522724867, "learning_rate": 3.2000000000000003e-06, "loss": 0.0, "num_tokens": 164320.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 65 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.010135135135135136, "frac_reward_zero_std": 1.0, "grad_norm": 6.449189822888002e-05, "kl": 0.0018529919907450676, "learning_rate": 3.2500000000000002e-06, "loss": 0.0, "num_tokens": 166848.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 66 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.010288697788697789, "frac_reward_zero_std": 0.0, "grad_norm": 0.32904744148254395, "kl": 0.001953460043296218, "learning_rate": 3.3000000000000006e-06, "loss": 0.0, "num_tokens": 169376.0, "reward": 0.5, "reward_std": 0.5345224738121033, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5345224738121033, "step": 67 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.010442260442260442, "frac_reward_zero_std": 0.0, "grad_norm": 0.281531423330307, "kl": 0.0020269351080060005, "learning_rate": 3.3500000000000005e-06, "loss": 0.0, "num_tokens": 171904.0, "reward": 0.5, "reward_std": 0.5345224738121033, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5345224738121033, "step": 68 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.010595823095823095, "frac_reward_zero_std": 1.0, "grad_norm": 5.164610774954781e-05, "kl": 0.0018625680822879076, "learning_rate": 3.4000000000000005e-06, "loss": 0.0, "num_tokens": 174432.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 69 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.01074938574938575, "frac_reward_zero_std": 0.0, "grad_norm": 0.2751414179801941, "kl": 0.0016131119336932898, "learning_rate": 3.45e-06, "loss": 0.0, "num_tokens": 176960.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 70 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.010902948402948403, "frac_reward_zero_std": 0.0, "grad_norm": 0.301176518201828, "kl": 0.0018663486698642373, "learning_rate": 3.5e-06, "loss": 0.0, "num_tokens": 179488.0, "reward": 0.625, "reward_std": 0.5175491571426392, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5175492167472839, "step": 71 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.011056511056511056, "frac_reward_zero_std": 1.0, "grad_norm": 4.354660995886661e-05, "kl": 0.0017401627264916897, "learning_rate": 3.5500000000000003e-06, "loss": 0.0, "num_tokens": 182016.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 72 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.01121007371007371, "frac_reward_zero_std": 1.0, "grad_norm": 5.4372500017052516e-05, "kl": 0.0021832087077200413, "learning_rate": 3.6000000000000003e-06, "loss": 0.0, "num_tokens": 184544.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 73 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.011363636363636364, "frac_reward_zero_std": 1.0, "grad_norm": 3.5604072763817385e-05, "kl": 0.0017031468451023102, "learning_rate": 3.65e-06, "loss": 0.0, "num_tokens": 187072.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 74 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.011517199017199017, "frac_reward_zero_std": 1.0, "grad_norm": 0.006764743477106094, "kl": 0.01521103922277689, "learning_rate": 3.7e-06, "loss": 0.0, "num_tokens": 189600.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 75 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.01167076167076167, "frac_reward_zero_std": 1.0, "grad_norm": 3.919710798072629e-05, "kl": 0.001588721526786685, "learning_rate": 3.7500000000000005e-06, "loss": 0.0, "num_tokens": 192128.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 76 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.011824324324324325, "frac_reward_zero_std": 0.0, "grad_norm": 0.283232718706131, "kl": 0.0022833647672086954, "learning_rate": 3.8000000000000005e-06, "loss": 0.0, "num_tokens": 194656.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 77 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.011977886977886978, "frac_reward_zero_std": 1.0, "grad_norm": 8.2063750596717e-05, "kl": 0.0021289209835231304, "learning_rate": 3.85e-06, "loss": 0.0, "num_tokens": 197184.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 78 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.012131449631449631, "frac_reward_zero_std": 0.0, "grad_norm": 0.23820413649082184, "kl": 0.0019389790249988437, "learning_rate": 3.900000000000001e-06, "loss": 0.0, "num_tokens": 199712.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 79 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.012285012285012284, "frac_reward_zero_std": 1.0, "grad_norm": 3.6227207601768896e-05, "kl": 0.0018583569908514619, "learning_rate": 3.95e-06, "loss": 0.0, "num_tokens": 202240.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 80 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.012438574938574939, "frac_reward_zero_std": 1.0, "grad_norm": 3.6503493902273476e-05, "kl": 0.0019133782479912043, "learning_rate": 4.000000000000001e-06, "loss": 0.0, "num_tokens": 204768.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 81 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.012592137592137592, "frac_reward_zero_std": 1.0, "grad_norm": 5.445562055683695e-05, "kl": 0.001825973391532898, "learning_rate": 4.05e-06, "loss": 0.0, "num_tokens": 207296.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 82 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.012745700245700245, "frac_reward_zero_std": 1.0, "grad_norm": 4.9896567361429334e-05, "kl": 0.0016772599192336202, "learning_rate": 4.1e-06, "loss": 0.0, "num_tokens": 209824.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 83 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.012899262899262898, "frac_reward_zero_std": 0.0, "grad_norm": 0.26036882400512695, "kl": 0.0018029706552624702, "learning_rate": 4.15e-06, "loss": 0.0, "num_tokens": 212352.0, "reward": 0.5, "reward_std": 0.5345224738121033, "rewards/compute_reward/mean": 0.5, "rewards/compute_reward/std": 0.5345224738121033, "step": 84 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.013052825552825553, "frac_reward_zero_std": 1.0, "grad_norm": 5.36849838681519e-05, "kl": 0.0016535220202058554, "learning_rate": 4.2000000000000004e-06, "loss": 0.0, "num_tokens": 214880.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 85 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.013206388206388206, "frac_reward_zero_std": 1.0, "grad_norm": 5.508507092599757e-05, "kl": 0.0019267231691628695, "learning_rate": 4.25e-06, "loss": 0.0, "num_tokens": 217408.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 86 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.01335995085995086, "frac_reward_zero_std": 1.0, "grad_norm": 3.793669384322129e-05, "kl": 0.0014971359632909298, "learning_rate": 4.3e-06, "loss": 0.0, "num_tokens": 219936.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 87 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.013513513513513514, "frac_reward_zero_std": 0.0, "grad_norm": 0.27244484424591064, "kl": 0.0018051433144137263, "learning_rate": 4.350000000000001e-06, "loss": 0.0, "num_tokens": 222464.0, "reward": 0.25, "reward_std": 0.4629100561141968, "rewards/compute_reward/mean": 0.25, "rewards/compute_reward/std": 0.4629100561141968, "step": 88 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.013667076167076167, "frac_reward_zero_std": 1.0, "grad_norm": 3.993747668573633e-05, "kl": 0.0018220169004052877, "learning_rate": 4.4e-06, "loss": 0.0, "num_tokens": 224992.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 89 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.01382063882063882, "frac_reward_zero_std": 1.0, "grad_norm": 8.624274050816894e-05, "kl": 0.0022467547096312046, "learning_rate": 4.450000000000001e-06, "loss": 0.0, "num_tokens": 227520.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 90 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.013974201474201474, "frac_reward_zero_std": 1.0, "grad_norm": 4.577819709083997e-05, "kl": 0.002002157736569643, "learning_rate": 4.5e-06, "loss": 0.0, "num_tokens": 230048.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 91 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.014127764127764128, "frac_reward_zero_std": 1.0, "grad_norm": 4.1842817154247314e-05, "kl": 0.0017484177369624376, "learning_rate": 4.5500000000000005e-06, "loss": 0.0, "num_tokens": 232576.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 92 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.014281326781326781, "frac_reward_zero_std": 0.0, "grad_norm": 0.2512575685977936, "kl": 0.0015832895878702402, "learning_rate": 4.600000000000001e-06, "loss": 0.0, "num_tokens": 235104.0, "reward": 0.625, "reward_std": 0.5175491571426392, "rewards/compute_reward/mean": 0.625, "rewards/compute_reward/std": 0.5175492167472839, "step": 93 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.014434889434889435, "frac_reward_zero_std": 1.0, "grad_norm": 7.139512308640406e-05, "kl": 0.0019443321507424116, "learning_rate": 4.65e-06, "loss": 0.0, "num_tokens": 237632.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 94 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.014588452088452088, "frac_reward_zero_std": 1.0, "grad_norm": 4.431122579262592e-05, "kl": 0.0016710343770682812, "learning_rate": 4.7e-06, "loss": 0.0, "num_tokens": 240160.0, "reward": 1.0, "reward_std": 0.0, "rewards/compute_reward/mean": 1.0, "rewards/compute_reward/std": 0.0, "step": 95 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.014742014742014743, "frac_reward_zero_std": 1.0, "grad_norm": 4.648687900044024e-05, "kl": 0.0017124009318649769, "learning_rate": 4.75e-06, "loss": 0.0, "num_tokens": 242688.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 96 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.014895577395577396, "frac_reward_zero_std": 1.0, "grad_norm": 5.7402332458877936e-05, "kl": 0.0018351635662838817, "learning_rate": 4.800000000000001e-06, "loss": 0.0, "num_tokens": 245216.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 97 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.015049140049140049, "frac_reward_zero_std": 0.0, "grad_norm": 0.2951975166797638, "kl": 0.001936566550284624, "learning_rate": 4.85e-06, "loss": 0.0, "num_tokens": 247744.0, "reward": 0.125, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.125, "rewards/compute_reward/std": 0.3535533845424652, "step": 98 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.015202702702702704, "frac_reward_zero_std": 1.0, "grad_norm": 3.6721663491334766e-05, "kl": 0.0014798862393945456, "learning_rate": 4.9000000000000005e-06, "loss": 0.0, "num_tokens": 250272.0, "reward": 0.0, "reward_std": 0.0, "rewards/compute_reward/mean": 0.0, "rewards/compute_reward/std": 0.0, "step": 99 }, { "completion_length": 256.0, "completions/clipped_ratio": 1.0, "completions/max_length": 256.0, "completions/max_terminated_length": 0.0, "completions/mean_length": 256.0, "completions/mean_terminated_length": 0.0, "completions/min_length": 256.0, "completions/min_terminated_length": 0.0, "epoch": 0.015356265356265357, "frac_reward_zero_std": 0.0, "grad_norm": 0.2804346978664398, "kl": 0.0018570123938843608, "learning_rate": 4.95e-06, "loss": 0.0, "num_tokens": 252800.0, "reward": 0.875, "reward_std": 0.3535533845424652, "rewards/compute_reward/mean": 0.875, "rewards/compute_reward/std": 0.3535533845424652, "step": 100 } ], "logging_steps": 1, "max_steps": 1000, "num_input_tokens_seen": 252800, "num_train_epochs": 1, "save_steps": 100, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }