| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.7293946024799417, |
| "global_step": 2000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.18, |
| "learning_rate": 4.817651349380015e-05, |
| "loss": 0.1396, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.36, |
| "learning_rate": 4.6353026987600294e-05, |
| "loss": 0.0182, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 4.452954048140044e-05, |
| "loss": 0.0148, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.73, |
| "learning_rate": 4.2706053975200585e-05, |
| "loss": 0.009, |
| "step": 2000 |
| } |
| ], |
| "max_steps": 13710, |
| "num_train_epochs": 5, |
| "total_flos": 304540228254096.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
