| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "global_step": 63, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.08, |
| "learning_rate": 0.00013507870183531476, |
| "loss": 3.734, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.16, |
| "learning_rate": 0.00012884599993319768, |
| "loss": 3.3859, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.24, |
| "learning_rate": 0.00011888735840752609, |
| "loss": 3.092, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.32, |
| "learning_rate": 0.0001058186737011911, |
| "loss": 3.5704, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.4, |
| "learning_rate": 9.044818420726556e-05, |
| "loss": 3.2569, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 7.372648442002871e-05, |
| "loss": 3.4692, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.56, |
| "learning_rate": 5.668773501204858e-05, |
| "loss": 3.1881, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.63, |
| "learning_rate": 4.0385704725240065e-05, |
| "loss": 3.1091, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.71, |
| "learning_rate": 2.5828599592490882e-05, |
| "loss": 3.1881, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.79, |
| "learning_rate": 1.3916710004507539e-05, |
| "loss": 2.9839, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.87, |
| "learning_rate": 5.38673186569003e-06, |
| "loss": 3.104, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.95, |
| "learning_rate": 7.662053209561833e-07, |
| "loss": 2.9588, |
| "step": 60 |
| } |
| ], |
| "max_steps": 63, |
| "num_train_epochs": 1, |
| "total_flos": 65714946048000.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
