| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.6856829402084477, |
| "global_step": 5000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.07, |
| "learning_rate": 2.897147558968733e-05, |
| "loss": 3.8538, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.14, |
| "learning_rate": 2.794295117937466e-05, |
| "loss": 2.9886, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.21, |
| "learning_rate": 2.6914426769061988e-05, |
| "loss": 2.685, |
| "step": 1500 |
| }, |
| { |
| "epoch": 0.27, |
| "learning_rate": 2.5885902358749317e-05, |
| "loss": 2.4911, |
| "step": 2000 |
| }, |
| { |
| "epoch": 0.34, |
| "learning_rate": 2.4857377948436646e-05, |
| "loss": 2.2862, |
| "step": 2500 |
| }, |
| { |
| "epoch": 0.41, |
| "learning_rate": 2.3828853538123975e-05, |
| "loss": 2.14, |
| "step": 3000 |
| }, |
| { |
| "epoch": 0.48, |
| "learning_rate": 2.28003291278113e-05, |
| "loss": 2.122, |
| "step": 3500 |
| }, |
| { |
| "epoch": 0.55, |
| "learning_rate": 2.177180471749863e-05, |
| "loss": 2.0886, |
| "step": 4000 |
| }, |
| { |
| "epoch": 0.62, |
| "learning_rate": 2.074328030718596e-05, |
| "loss": 2.0471, |
| "step": 4500 |
| }, |
| { |
| "epoch": 0.69, |
| "learning_rate": 1.9714755896873284e-05, |
| "loss": 1.9984, |
| "step": 5000 |
| } |
| ], |
| "max_steps": 14584, |
| "num_train_epochs": 2, |
| "total_flos": 631754864640000.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
