{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 0.4, "eval_steps": 500, "global_step": 60, "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": 0.4125, "completions/max_length": 512.0, "completions/max_terminated_length": 489.8, "completions/mean_length": 459.8, "completions/mean_terminated_length": 425.1761993408203, "completions/min_length": 345.6, "completions/min_terminated_length": 345.6, "entropy": 0.4540576159954071, "epoch": 0.06666666666666667, "frac_reward_zero_std": 0.15, "grad_norm": 2.28125, "kl": 0.02153960674040718, "learning_rate": 9.100000000000001e-06, "loss": 0.019999520480632783, "num_tokens": 49652.0, "reward": 0.855133056640625, "reward_std": 0.38260793387889863, "rewards/JointRewardFunction/mean": 0.855133056640625, "rewards/JointRewardFunction/std": 0.3826079457998276, "step": 10, "step_time": 22.616553169000053 }, { "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": 0.325, "completions/max_length": 502.0, "completions/max_terminated_length": 441.3, "completions/mean_length": 422.4875, "completions/mean_terminated_length": 381.8863128662109, "completions/min_length": 320.2, "completions/min_terminated_length": 320.2, "entropy": 0.49893306717276575, "epoch": 0.13333333333333333, "frac_reward_zero_std": 0.5, "grad_norm": 1.8828125, "kl": 0.030561967543326317, "learning_rate": 8.1e-06, "loss": 0.015025660395622253, "num_tokens": 97211.0, "reward": 0.925, "reward_std": 0.4372564494609833, "rewards/JointRewardFunction/mean": 0.925, "rewards/JointRewardFunction/std": 0.4372564613819122, "step": 20, "step_time": 21.99016141649995 }, { "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": 0.1875, "completions/max_length": 482.7, "completions/max_terminated_length": 447.1, "completions/mean_length": 385.025, "completions/mean_terminated_length": 360.42596130371095, "completions/min_length": 259.8, "completions/min_terminated_length": 259.8, "entropy": 0.5374923676252366, "epoch": 0.2, "frac_reward_zero_std": 0.3, "grad_norm": 2.03125, "kl": 0.054876458749640734, "learning_rate": 7.100000000000001e-06, "loss": 0.02556714415550232, "num_tokens": 139429.0, "reward": 1.0125, "reward_std": 0.40190111100673676, "rewards/JointRewardFunction/mean": 1.0125, "rewards/JointRewardFunction/std": 0.4019011229276657, "step": 30, "step_time": 21.189481891999776 }, { "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": 0.3625, "completions/max_length": 508.6, "completions/max_terminated_length": 418.1, "completions/mean_length": 419.1125, "completions/mean_terminated_length": 336.9939331054687, "completions/min_length": 313.1, "completions/min_terminated_length": 261.9, "entropy": 0.4931879937648773, "epoch": 0.26666666666666666, "frac_reward_zero_std": 0.35, "grad_norm": 0.00390625, "kl": 0.049289348628371955, "learning_rate": 6.1e-06, "loss": 0.008195620030164719, "num_tokens": 186270.0, "reward": 0.9625, "reward_std": 0.3863321661949158, "rewards/JointRewardFunction/mean": 0.9625, "rewards/JointRewardFunction/std": 0.3863321840763092, "step": 40, "step_time": 22.299921856999937 }, { "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": 0.2125, "completions/max_length": 502.5, "completions/max_terminated_length": 466.4, "completions/mean_length": 407.3, "completions/mean_terminated_length": 383.09393920898435, "completions/min_length": 315.2, "completions/min_terminated_length": 315.2, "entropy": 0.5289955049753189, "epoch": 0.3333333333333333, "frac_reward_zero_std": 0.45, "grad_norm": 3.125, "kl": 0.04389781395439059, "learning_rate": 5.1e-06, "loss": 0.03555458188056946, "num_tokens": 231570.0, "reward": 1.0125, "reward_std": 0.3774814397096634, "rewards/JointRewardFunction/mean": 1.0125, "rewards/JointRewardFunction/std": 0.3774814516305923, "step": 50, "step_time": 22.136872237800254 }, { "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": 0.15, "completions/max_length": 492.1, "completions/max_terminated_length": 475.5, "completions/mean_length": 396.575, "completions/mean_terminated_length": 380.23631591796874, "completions/min_length": 299.5, "completions/min_terminated_length": 299.5, "entropy": 0.47242984138429167, "epoch": 0.4, "frac_reward_zero_std": 0.5, "grad_norm": 1.78125, "kl": 0.050327780924271794, "learning_rate": 4.1e-06, "loss": -0.009574128687381745, "num_tokens": 276696.0, "reward": 1.0625, "reward_std": 0.36363019943237307, "rewards/JointRewardFunction/mean": 1.0625, "rewards/JointRewardFunction/std": 0.3636302053928375, "step": 60, "step_time": 21.720717019600034 } ], "logging_steps": 10, "max_steps": 100, "num_input_tokens_seen": 276696, "num_train_epochs": 1, "save_steps": 20, "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": 1, "trial_name": null, "trial_params": null }