| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 6.0, | |
| "global_step": 3834, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.793975255513717e-05, | |
| "loss": 0.0241, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.5250134480903713e-05, | |
| "loss": 0.0021, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.2560516406670254e-05, | |
| "loss": 0.0014, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 9.870898332436795e-06, | |
| "loss": 0.0011, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 7.181280258203336e-06, | |
| "loss": 0.0008, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 4.491662183969877e-06, | |
| "loss": 0.0006, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 5.48, | |
| "learning_rate": 1.8020441097364175e-06, | |
| "loss": 0.0005, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "step": 3834, | |
| "total_flos": 2403004334473216.0, | |
| "train_loss": 0.0040491660264984795, | |
| "train_runtime": 24808.6685, | |
| "train_samples_per_second": 19.777, | |
| "train_steps_per_second": 0.155 | |
| } | |
| ], | |
| "max_steps": 3834, | |
| "num_train_epochs": 6, | |
| "total_flos": 2403004334473216.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |