| { | |
| "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": 3.3748, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9161849710982662e-05, | |
| "loss": 2.6048, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 42.999053926206244, | |
| "eval_f1": 54.526276703809444, | |
| "step": 1384 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.3742774566473989e-05, | |
| "loss": 2.3908, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 8.323699421965318e-06, | |
| "loss": 2.2263, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 2.9046242774566473e-06, | |
| "loss": 2.1578, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 47.60643330179754, | |
| "eval_f1": 58.95492494932076, | |
| "step": 2768 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 2768, | |
| "total_flos": 1289049683787776.0, | |
| "train_loss": 2.512806776631085, | |
| "train_runtime": 380.1872, | |
| "train_samples_per_second": 465.686, | |
| "train_steps_per_second": 7.281 | |
| } | |
| ], | |
| "max_steps": 2768, | |
| "num_train_epochs": 2, | |
| "total_flos": 1289049683787776.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |