| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "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": 351.71875, | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 6.811317443847656, | |
| "kl": 0.0, | |
| "learning_rate": 5e-07, | |
| "loss": 0.0, | |
| "reward": 1.785135745536536, | |
| "reward_std": 0.7207677089609206, | |
| "rewards/concensus_correctness_reward_func": 0.21193749643862247, | |
| "rewards/consensus_reward_func": 0.4375, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.5477294897427782, | |
| "rewards/soft_format_reward_func": 0.015625, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.5723437489941716, | |
| "step": 2 | |
| }, | |
| { | |
| "completion_length": 300.7083333333333, | |
| "epoch": 1.0, | |
| "grad_norm": 4.019150733947754, | |
| "kl": 0.0015880058393425618, | |
| "learning_rate": 4.864543104251586e-07, | |
| "loss": 0.0, | |
| "reward": 2.815759986639023, | |
| "reward_std": 1.311702584226926, | |
| "rewards/concensus_correctness_reward_func": 0.6080833251277605, | |
| "rewards/consensus_reward_func": 0.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.5803016399343809, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.020833333333333332, | |
| "rewards/xmlcount_reward_func": 0.8565416683753332, | |
| "step": 4 | |
| }, | |
| { | |
| "completion_length": 267.96875, | |
| "epoch": 1.5714285714285714, | |
| "grad_norm": 6.299780368804932, | |
| "kl": 0.0036450710395001806, | |
| "learning_rate": 4.472851273490984e-07, | |
| "loss": 0.0, | |
| "reward": 2.7576604560017586, | |
| "reward_std": 1.438764474238269, | |
| "rewards/concensus_correctness_reward_func": 0.6950624939054251, | |
| "rewards/consensus_reward_func": 0.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.5075979293324053, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.015625, | |
| "rewards/xmlcount_reward_func": 0.7893750090152025, | |
| "step": 6 | |
| }, | |
| { | |
| "completion_length": 263.75, | |
| "epoch": 2.0, | |
| "grad_norm": 4.339041709899902, | |
| "kl": 0.005562390239598851, | |
| "learning_rate": 3.867370395306068e-07, | |
| "loss": 0.0, | |
| "reward": 2.794683963060379, | |
| "reward_std": 1.0713556228826444, | |
| "rewards/concensus_correctness_reward_func": 0.5786666596929232, | |
| "rewards/consensus_reward_func": 1.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.5789339219530424, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.020833333333333332, | |
| "rewards/xmlcount_reward_func": 0.6162500108281771, | |
| "step": 8 | |
| }, | |
| { | |
| "completion_length": 219.1875, | |
| "epoch": 2.571428571428571, | |
| "grad_norm": 7.3506574630737305, | |
| "kl": 0.03309335559606552, | |
| "learning_rate": 3.1137137178519977e-07, | |
| "loss": 0.0, | |
| "reward": 3.0435366928577423, | |
| "reward_std": 1.0485685534076765, | |
| "rewards/concensus_correctness_reward_func": 0.6486249938607216, | |
| "rewards/consensus_reward_func": 1.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.48016168363392353, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.015625, | |
| "rewards/xmlcount_reward_func": 0.899125000461936, | |
| "step": 10 | |
| }, | |
| { | |
| "completion_length": 318.7916666666667, | |
| "epoch": 3.0, | |
| "grad_norm": 3.507310390472412, | |
| "kl": 0.007355922580851863, | |
| "learning_rate": 2.2935516363191693e-07, | |
| "loss": 0.0, | |
| "reward": 3.699404043455919, | |
| "reward_std": 1.2840088397885363, | |
| "rewards/concensus_correctness_reward_func": 1.0116666505734127, | |
| "rewards/consensus_reward_func": 1.4166666666666667, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.6990290557344755, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.020833333333333332, | |
| "rewards/xmlcount_reward_func": 0.5512083147962888, | |
| "step": 12 | |
| }, | |
| { | |
| "completion_length": 298.1875, | |
| "epoch": 3.571428571428571, | |
| "grad_norm": 6.966953754425049, | |
| "kl": 0.013355173650779761, | |
| "learning_rate": 1.4957614383675767e-07, | |
| "loss": 0.0, | |
| "reward": 2.5207905806601048, | |
| "reward_std": 1.5945063475519419, | |
| "rewards/concensus_correctness_reward_func": 0.508437491953373, | |
| "rewards/consensus_reward_func": 0.75, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.5373843046836555, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.015625, | |
| "rewards/xmlcount_reward_func": 0.7093437574803829, | |
| "step": 14 | |
| }, | |
| { | |
| "completion_length": 243.0, | |
| "epoch": 4.0, | |
| "grad_norm": 5.53330659866333, | |
| "kl": 0.015268806600943208, | |
| "learning_rate": 8.067960709356478e-08, | |
| "loss": 0.0, | |
| "reward": 1.8712952236334484, | |
| "reward_std": 1.110026717806856, | |
| "rewards/concensus_correctness_reward_func": 0.250750000278155, | |
| "rewards/consensus_reward_func": 0.4166666666666667, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.4589619180187583, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.7449166675408682, | |
| "step": 16 | |
| }, | |
| { | |
| "completion_length": 381.1875, | |
| "epoch": 4.571428571428571, | |
| "grad_norm": 6.215991020202637, | |
| "kl": 0.00821731197356712, | |
| "learning_rate": 3.013156219837776e-08, | |
| "loss": 0.0, | |
| "reward": 2.828702801838517, | |
| "reward_std": 1.5806347676552832, | |
| "rewards/concensus_correctness_reward_func": 0.5827499907463789, | |
| "rewards/consensus_reward_func": 1.0, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.47042151528876275, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.015625, | |
| "rewards/xmlcount_reward_func": 0.7599062472581863, | |
| "step": 18 | |
| }, | |
| { | |
| "completion_length": 334.0416666666667, | |
| "epoch": 5.0, | |
| "grad_norm": 4.492783069610596, | |
| "kl": 0.00856334783990557, | |
| "learning_rate": 3.4096741493194193e-09, | |
| "loss": 0.0, | |
| "reward": 2.5465123057365417, | |
| "reward_std": 1.2983978067835171, | |
| "rewards/concensus_correctness_reward_func": 0.6519999926288923, | |
| "rewards/consensus_reward_func": 0.8333333333333334, | |
| "rewards/cumulative_reward_2": 0.0, | |
| "rewards/final_correctness_reward_func": 0.0, | |
| "rewards/question_recreation_reward_func": 0.46867894288152456, | |
| "rewards/soft_format_reward_func": 0.0, | |
| "rewards/strict_format_reward_func": 0.0, | |
| "rewards/xmlcount_reward_func": 0.5925000111262003, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 20, | |
| "total_flos": 0.0, | |
| "train_loss": 8.644163608551026e-06, | |
| "train_runtime": 197.5707, | |
| "train_samples_per_second": 1.62, | |
| "train_steps_per_second": 0.101 | |
| } | |
| ], | |
| "logging_steps": 2, | |
| "max_steps": 20, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 7, | |
| "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 | |
| } | |