| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.1881838074398248, |
| "global_step": 6000, |
| "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 |
| }, |
| { |
| "epoch": 0.91, |
| "learning_rate": 4.088256746900073e-05, |
| "loss": 0.0083, |
| "step": 2500 |
| }, |
| { |
| "epoch": 1.09, |
| "learning_rate": 3.9059080962800876e-05, |
| "loss": 0.0072, |
| "step": 3000 |
| }, |
| { |
| "epoch": 1.28, |
| "learning_rate": 3.723559445660103e-05, |
| "loss": 0.004, |
| "step": 3500 |
| }, |
| { |
| "epoch": 1.46, |
| "learning_rate": 3.541210795040117e-05, |
| "loss": 0.0052, |
| "step": 4000 |
| }, |
| { |
| "epoch": 1.64, |
| "learning_rate": 3.358862144420131e-05, |
| "loss": 0.0032, |
| "step": 4500 |
| }, |
| { |
| "epoch": 1.82, |
| "learning_rate": 3.1765134938001465e-05, |
| "loss": 0.0029, |
| "step": 5000 |
| }, |
| { |
| "epoch": 2.01, |
| "learning_rate": 2.9941648431801604e-05, |
| "loss": 0.0055, |
| "step": 5500 |
| }, |
| { |
| "epoch": 2.19, |
| "learning_rate": 2.811816192560175e-05, |
| "loss": 0.0027, |
| "step": 6000 |
| } |
| ], |
| "max_steps": 13710, |
| "num_train_epochs": 5, |
| "total_flos": 913869748994976.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
