| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 2768, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 2.458092485549133e-05, | |
| "loss": 4.4875, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9161849710982662e-05, | |
| "loss": 3.9584, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 14.862819299905393, | |
| "eval_f1": 26.85226111240762, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.3742774566473989e-05, | |
| "loss": 3.7147, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.323699421965318e-06, | |
| "loss": 3.5194, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.9046242774566473e-06, | |
| "loss": 3.4412, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 20.53926206244087, | |
| "eval_f1": 32.48117986953082, | |
| "step": 2768 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 2768, | |
| "total_flos": 161967338160128.0, | |
| "train_loss": 3.785540608312353, | |
| "train_runtime": 234.5103, | |
| "train_samples_per_second": 754.969, | |
| "train_steps_per_second": 11.803 | |
| } | |
| ], | |
| "max_steps": 2768, | |
| "num_train_epochs": 2, | |
| "total_flos": 161967338160128.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |