| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.7619047619047619, | |
| "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": 148.84375, | |
| "epoch": 0.0761904761904762, | |
| "grad_norm": 68.10345458984375, | |
| "kl": 0.0, | |
| "learning_rate": 4.965903258506806e-07, | |
| "loss": -0.0, | |
| "reward": 0.10526368767023087, | |
| "reward_std": 0.12028272116640437, | |
| "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": 0.049951187102124095, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": -0.0071875001303851604, | |
| "step": 2 | |
| }, | |
| { | |
| "completion_length": 139.75, | |
| "epoch": 0.1523809523809524, | |
| "grad_norm": 109.3356704711914, | |
| "kl": 0.010293949244214673, | |
| "learning_rate": 4.698684378016222e-07, | |
| "loss": 0.0, | |
| "reward": 0.1607844315294642, | |
| "reward_std": 0.13128481482999632, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.09828442826983519, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 4 | |
| }, | |
| { | |
| "completion_length": 215.03125, | |
| "epoch": 0.22857142857142856, | |
| "grad_norm": 39.21604537963867, | |
| "kl": 0.009948016217776967, | |
| "learning_rate": 4.193203929064353e-07, | |
| "loss": 0.0, | |
| "reward": 0.047717132983962074, | |
| "reward_std": 0.02130908235994866, | |
| "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": 0.047717132983962074, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 6 | |
| }, | |
| { | |
| "completion_length": 88.65625, | |
| "epoch": 0.3047619047619048, | |
| "grad_norm": 9.36317253112793, | |
| "kl": 0.009089788440178381, | |
| "learning_rate": 3.5042385616324236e-07, | |
| "loss": 0.0, | |
| "reward": 0.07478685077512637, | |
| "reward_std": 0.027659101046310752, | |
| "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": 0.07478685077512637, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 8 | |
| }, | |
| { | |
| "completion_length": 180.1875, | |
| "epoch": 0.38095238095238093, | |
| "grad_norm": 17.74615478515625, | |
| "kl": 0.004952711342411931, | |
| "learning_rate": 2.706448363680831e-07, | |
| "loss": 0.0, | |
| "reward": 0.1271270774595905, | |
| "reward_std": 0.11367859682104609, | |
| "rewards/concensus_correctness_reward_func": 0.0, | |
| "rewards/consensus_reward_func": 0.0625, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.06056457760860212, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.004062499850988388, | |
| "step": 10 | |
| }, | |
| { | |
| "completion_length": 244.4375, | |
| "epoch": 0.45714285714285713, | |
| "grad_norm": 10.840444564819336, | |
| "kl": 0.009140046095126309, | |
| "learning_rate": 1.886286282148002e-07, | |
| "loss": 0.0, | |
| "reward": 0.07606052907067351, | |
| "reward_std": 0.04101635366532719, | |
| "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": 0.07606052907067351, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 12 | |
| }, | |
| { | |
| "completion_length": 175.40625, | |
| "epoch": 0.5333333333333333, | |
| "grad_norm": 28.949663162231445, | |
| "kl": 0.010847578349057585, | |
| "learning_rate": 1.1326296046939333e-07, | |
| "loss": 0.0, | |
| "reward": 0.056656608707271516, | |
| "reward_std": 0.03052737117104698, | |
| "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": 0.056656608707271516, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 14 | |
| }, | |
| { | |
| "completion_length": 174.375, | |
| "epoch": 0.6095238095238096, | |
| "grad_norm": 8.086259841918945, | |
| "kl": 0.017883143130802637, | |
| "learning_rate": 5.271487265090163e-08, | |
| "loss": 0.0, | |
| "reward": 0.05451725912280381, | |
| "reward_std": 0.020318549079092918, | |
| "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": 0.05451725912280381, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 16 | |
| }, | |
| { | |
| "completion_length": 212.78125, | |
| "epoch": 0.6857142857142857, | |
| "grad_norm": 9.846426010131836, | |
| "kl": 0.013088275671179872, | |
| "learning_rate": 1.3545689574841341e-08, | |
| "loss": 0.0, | |
| "reward": 0.20598165242699906, | |
| "reward_std": 0.2135821714764461, | |
| "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": 0.08098165650153533, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.0, | |
| "step": 18 | |
| }, | |
| { | |
| "completion_length": 295.40625, | |
| "epoch": 0.7619047619047619, | |
| "grad_norm": 26.243383407592773, | |
| "kl": 0.0039032802706060465, | |
| "learning_rate": 0.0, | |
| "loss": 0.0, | |
| "reward": 0.1307674559066072, | |
| "reward_std": 0.15399234421056462, | |
| "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": 0.08804870548192412, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": -0.01978125050663948, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.7619047619047619, | |
| "step": 20, | |
| "total_flos": 0.0, | |
| "train_loss": 8.890809567674296e-06, | |
| "train_runtime": 12059.5712, | |
| "train_samples_per_second": 0.027, | |
| "train_steps_per_second": 0.002 | |
| } | |
| ], | |
| "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 | |
| } | |