| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.5, | |
| "eval_steps": 500, | |
| "global_step": 20, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "completion_length": 282.71875, | |
| "epoch": 0.05, | |
| "grad_norm": 42.93820571899414, | |
| "kl": 0.0, | |
| "learning_rate": 4.965903258506806e-07, | |
| "loss": 0.0, | |
| "reward": 0.6292939200648107, | |
| "reward_std": 0.6506021634559147, | |
| "rewards/concensus_correctness_reward_func": 0.04662499949336052, | |
| "rewards/consensus_reward_func": 0.0625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.25, | |
| "rewards/question_recreation_reward_func": 0.14407516870414838, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.1260937498882413, | |
| "step": 2 | |
| }, | |
| { | |
| "completion_length": 316.15625, | |
| "epoch": 0.1, | |
| "grad_norm": 11.557947158813477, | |
| "kl": 0.002068059044177062, | |
| "learning_rate": 4.698684378016222e-07, | |
| "loss": 0.0, | |
| "reward": 0.5142579441308044, | |
| "reward_std": 0.40489116005483083, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.1875, | |
| "rewards/question_recreation_reward_func": 0.22807045001536608, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.09868749883025885, | |
| "step": 4 | |
| }, | |
| { | |
| "completion_length": 397.90625, | |
| "epoch": 0.15, | |
| "grad_norm": 33.66463851928711, | |
| "kl": 0.0023345137960859574, | |
| "learning_rate": 4.193203929064353e-07, | |
| "loss": 0.0, | |
| "reward": 0.0876373503706418, | |
| "reward_std": 0.2730903436749941, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.12148110551061109, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": -0.0338437519967556, | |
| "step": 6 | |
| }, | |
| { | |
| "completion_length": 322.21875, | |
| "epoch": 0.2, | |
| "grad_norm": 13.174933433532715, | |
| "kl": 0.0018492022645659745, | |
| "learning_rate": 3.5042385616324236e-07, | |
| "loss": 0.0, | |
| "reward": 0.41745682479813695, | |
| "reward_std": 0.3582948070834391, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.125, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0625, | |
| "rewards/question_recreation_reward_func": 0.1773943287844304, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.052562499884516, | |
| "step": 8 | |
| }, | |
| { | |
| "completion_length": 288.125, | |
| "epoch": 0.25, | |
| "grad_norm": 67.86888122558594, | |
| "kl": 0.008549842281354358, | |
| "learning_rate": 2.706448363680831e-07, | |
| "loss": 0.0, | |
| "reward": 0.40863744184025563, | |
| "reward_std": 0.49519592405704316, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0625, | |
| "rewards/question_recreation_reward_func": 0.1976999432372395, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.015625, | |
| "rewards/xmlcount_reward_func": 0.07031250046566129, | |
| "step": 10 | |
| }, | |
| { | |
| "completion_length": 212.03125, | |
| "epoch": 0.3, | |
| "grad_norm": 24.831544876098633, | |
| "kl": 0.008816897239739774, | |
| "learning_rate": 1.886286282148002e-07, | |
| "loss": 0.0, | |
| "reward": 0.36399466946022585, | |
| "reward_std": 0.2636154612409882, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.18833842262392864, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.17565625021234155, | |
| "step": 12 | |
| }, | |
| { | |
| "completion_length": 369.34375, | |
| "epoch": 0.35, | |
| "grad_norm": 4.622942924499512, | |
| "kl": 0.002492460354915238, | |
| "learning_rate": 1.1326296046939333e-07, | |
| "loss": 0.0, | |
| "reward": 0.13834899943321943, | |
| "reward_std": 0.13620900230307598, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.07897400224464945, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.059374999604187906, | |
| "step": 14 | |
| }, | |
| { | |
| "completion_length": 334.78125, | |
| "epoch": 0.4, | |
| "grad_norm": 42.03730010986328, | |
| "kl": 0.005604283696811763, | |
| "learning_rate": 5.271487265090163e-08, | |
| "loss": 0.0, | |
| "reward": 1.140276842750609, | |
| "reward_std": 0.9167491418556892, | |
| "rewards/concensus_correctness_reward_func": 0.7786874994635582, | |
| "rewards/consensus_reward_func": 0.0625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.125, | |
| "rewards/question_recreation_reward_func": 0.11102682136697695, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.06306249997578561, | |
| "step": 16 | |
| }, | |
| { | |
| "completion_length": 370.53125, | |
| "epoch": 0.45, | |
| "grad_norm": 103.69042205810547, | |
| "kl": 0.016677564308338333, | |
| "learning_rate": 1.3545689574841341e-08, | |
| "loss": 0.0, | |
| "reward": 0.251421230728738, | |
| "reward_std": 0.507526803878136, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0625, | |
| "rewards/question_recreation_reward_func": 0.2050462217302993, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": -0.01612500031478703, | |
| "step": 18 | |
| }, | |
| { | |
| "completion_length": 309.03125, | |
| "epoch": 0.5, | |
| "grad_norm": 18.9988956451416, | |
| "kl": 0.0024696916407265235, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "reward": 0.4606377884047106, | |
| "reward_std": 0.46051671362511115, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.125, | |
| "rewards/question_recreation_reward_func": 0.16510653705336154, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.015625, | |
| "rewards/xmlcount_reward_func": 0.1549062510021031, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "step": 20, | |
| "total_flos": 0.0, | |
| "train_loss": 5.050748984558595e-06, | |
| "train_runtime": 778.5481, | |
| "train_samples_per_second": 0.411, | |
| "train_steps_per_second": 0.026 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 20, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 25, | |
| "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 | |
| } | |