{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9906191369606003, "eval_steps": 100, "global_step": 66, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "clip_ratio": 0.0, "completion_length": 676.4062728881836, "epoch": 0.0150093808630394, "grad_norm": 0.47017931938171387, "kl": 0.0, "learning_rate": 4.2857142857142857e-07, "loss": 0.0201, "reward": 0.4944196678698063, "reward_std": 0.3951864466071129, "rewards/accuracy_reward": 0.4810268171131611, "rewards/format_reward": 0.01339285762514919, "step": 1 }, { "clip_ratio": 0.0, "completion_length": 653.8295516967773, "epoch": 0.075046904315197, "grad_norm": 1.3977434635162354, "kl": 0.00030475854873657227, "learning_rate": 2.142857142857143e-06, "loss": 0.0373, "reward": 0.48604912869632244, "reward_std": 0.4064541347324848, "rewards/accuracy_reward": 0.47014511190354824, "rewards/format_reward": 0.01590401868452318, "step": 5 }, { "clip_ratio": 0.0, "completion_length": 680.8995895385742, "epoch": 0.150093808630394, "grad_norm": 49.64258575439453, "kl": 0.029695892333984376, "learning_rate": 2.980902532082017e-06, "loss": 0.0386, "reward": 0.5100446656346321, "reward_std": 0.3609618041664362, "rewards/accuracy_reward": 0.49888395220041276, "rewards/format_reward": 0.011160714947618545, "step": 10 }, { "clip_ratio": 0.0, "completion_length": 672.1567276000976, "epoch": 0.225140712945591, "grad_norm": 42.10508728027344, "kl": 0.0667633056640625, "learning_rate": 2.8659521592823702e-06, "loss": 0.0692, "reward": 0.6156250298023224, "reward_std": 0.3105337034910917, "rewards/accuracy_reward": 0.6107143193483353, "rewards/format_reward": 0.004910714528523386, "step": 15 }, { "clip_ratio": 0.0, "completion_length": 669.3482452392578, "epoch": 0.300187617260788, "grad_norm": 0.28610214591026306, "kl": 0.012540817260742188, "learning_rate": 2.6547509751448593e-06, "loss": 0.0597, "reward": 0.6343750298023224, "reward_std": 0.29044545218348505, "rewards/accuracy_reward": 0.6227678835391999, "rewards/format_reward": 0.011607143469154835, "step": 20 }, { "clip_ratio": 0.0, "completion_length": 657.5980209350586, "epoch": 0.37523452157598497, "grad_norm": 0.27337566018104553, "kl": 0.015349578857421876, "learning_rate": 2.3621811153216106e-06, "loss": 0.0454, "reward": 0.6906250327825546, "reward_std": 0.35538587234914304, "rewards/accuracy_reward": 0.6098214581608772, "rewards/format_reward": 0.08080357492435723, "step": 25 }, { "clip_ratio": 0.0, "completion_length": 614.5393112182617, "epoch": 0.450281425891182, "grad_norm": 0.32892781496047974, "kl": 0.04139404296875, "learning_rate": 2.008858299177045e-06, "loss": 0.0704, "reward": 0.8767857551574707, "reward_std": 0.5013510406017303, "rewards/accuracy_reward": 0.5830357387661934, "rewards/format_reward": 0.2937500152736902, "step": 30 }, { "clip_ratio": 0.0, "completion_length": 595.8272560119628, "epoch": 0.525328330206379, "grad_norm": 0.376099556684494, "kl": 0.0513580322265625, "learning_rate": 1.6196791583296247e-06, "loss": 0.0765, "reward": 0.9892857626080513, "reward_std": 0.5466332092881203, "rewards/accuracy_reward": 0.6133928909897804, "rewards/format_reward": 0.37589287385344505, "step": 35 }, { "clip_ratio": 0.0, "completion_length": 589.1964538574218, "epoch": 0.600375234521576, "grad_norm": 1.8511918783187866, "kl": 0.06317138671875, "learning_rate": 1.2220669138693288e-06, "loss": 0.0653, "reward": 0.9149554014205933, "reward_std": 0.5478003636002541, "rewards/accuracy_reward": 0.5924107439815998, "rewards/format_reward": 0.3225446578115225, "step": 40 }, { "clip_ratio": 0.0, "completion_length": 569.7413261413574, "epoch": 0.6754221388367729, "grad_norm": 0.9464062452316284, "kl": 0.083282470703125, "learning_rate": 8.44039019311717e-07, "loss": 0.063, "reward": 0.9430804029107094, "reward_std": 0.5471669748425484, "rewards/accuracy_reward": 0.6002232380211353, "rewards/format_reward": 0.34285715855658055, "step": 45 }, { "clip_ratio": 0.0, "completion_length": 568.6540458679199, "epoch": 0.7504690431519699, "grad_norm": 0.6890137195587158, "kl": 0.1210205078125, "learning_rate": 5.122329314024422e-07, "loss": 0.0644, "reward": 1.0017857611179353, "reward_std": 0.5992289334535599, "rewards/accuracy_reward": 0.5618303835391998, "rewards/format_reward": 0.43995537608861923, "step": 50 }, { "clip_ratio": 0.0, "completion_length": 565.0426589965821, "epoch": 0.8255159474671669, "grad_norm": 0.7971926927566528, "kl": 0.1895751953125, "learning_rate": 2.5002912137622743e-07, "loss": 0.0631, "reward": 0.9792411208152771, "reward_std": 0.6061119914054871, "rewards/accuracy_reward": 0.5354910954833031, "rewards/format_reward": 0.44375001862645147, "step": 55 }, { "clip_ratio": 0.0, "completion_length": 559.029933166504, "epoch": 0.900562851782364, "grad_norm": 6.136556625366211, "kl": 0.21927490234375, "learning_rate": 7.590358729742808e-08, "loss": 0.0585, "reward": 0.9495536148548126, "reward_std": 0.6090080454945564, "rewards/accuracy_reward": 0.5138393141329288, "rewards/format_reward": 0.4357143059372902, "step": 60 }, { "clip_ratio": 0.0, "completion_length": 545.569448852539, "epoch": 0.975609756097561, "grad_norm": 2.442777395248413, "kl": 0.2308837890625, "learning_rate": 2.1259564848570834e-09, "loss": 0.0727, "reward": 0.9575893312692643, "reward_std": 0.6195845372974873, "rewards/accuracy_reward": 0.5162946693599224, "rewards/format_reward": 0.4412946619093418, "step": 65 }, { "clip_ratio": 0.0, "completion_length": 596.2159271240234, "epoch": 0.9906191369606003, "kl": 0.2452392578125, "reward": 0.8973214700818062, "reward_std": 0.619016133248806, "rewards/accuracy_reward": 0.4754464514553547, "rewards/format_reward": 0.4218750149011612, "step": 66, "total_flos": 0.0, "train_loss": 0.060763326275980835, "train_runtime": 7634.6285, "train_samples_per_second": 1.116, "train_steps_per_second": 0.009 } ], "logging_steps": 5, "max_steps": 66, "num_input_tokens_seen": 0, "num_train_epochs": 1, "save_steps": 500, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": false, "should_training_stop": false }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 16, "trial_name": null, "trial_params": null }