| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 2290, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 3.9082969432314415e-05, | |
| "loss": 1.7186, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 2.816593886462882e-05, | |
| "loss": 1.2843, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 1.7248908296943234e-05, | |
| "loss": 1.189, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 6.3318777292576415e-06, | |
| "loss": 1.1345, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 2290, | |
| "total_flos": 3.165553346892595e+16, | |
| "train_loss": 1.3039255067250615, | |
| "train_runtime": 1845.507, | |
| "train_samples_per_second": 12.392, | |
| "train_steps_per_second": 1.241 | |
| } | |
| ], | |
| "max_steps": 2290, | |
| "num_train_epochs": 10, | |
| "total_flos": 3.165553346892595e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |