| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 3.0, |
| "global_step": 4389, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.34, |
| "learning_rate": 5.1274204815916455e-05, |
| "loss": 1.8883, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.68, |
| "learning_rate": 4.4719077634132206e-05, |
| "loss": 1.1957, |
| "step": 1000 |
| }, |
| { |
| "epoch": 1.03, |
| "learning_rate": 3.816395045234795e-05, |
| "loss": 1.071, |
| "step": 1500 |
| }, |
| { |
| "epoch": 1.37, |
| "learning_rate": 3.1608823270563686e-05, |
| "loss": 0.8373, |
| "step": 2000 |
| }, |
| { |
| "epoch": 1.71, |
| "learning_rate": 2.5066806343143003e-05, |
| "loss": 0.8106, |
| "step": 2500 |
| }, |
| { |
| "epoch": 2.05, |
| "learning_rate": 1.8511679161358743e-05, |
| "loss": 0.7633, |
| "step": 3000 |
| }, |
| { |
| "epoch": 2.39, |
| "learning_rate": 1.1956551979574485e-05, |
| "loss": 0.5825, |
| "step": 3500 |
| }, |
| { |
| "epoch": 2.73, |
| "learning_rate": 5.401424797790227e-06, |
| "loss": 0.5819, |
| "step": 4000 |
| }, |
| { |
| "epoch": 3.0, |
| "step": 4389, |
| "total_flos": 62595188391936.0, |
| "train_loss": 0.9303430125718237, |
| "train_runtime": 645.7605, |
| "train_samples_per_second": 407.589, |
| "train_steps_per_second": 6.797 |
| } |
| ], |
| "max_steps": 4389, |
| "num_train_epochs": 3, |
| "total_flos": 62595188391936.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
