| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 2100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 3.809523809523809e-05, | |
| "loss": 7.339, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 2.6190476190476192e-05, | |
| "loss": 4.7056, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 1.4285714285714285e-05, | |
| "loss": 1.4841, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 2.3809523809523808e-06, | |
| "loss": 0.563, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 2100, | |
| "total_flos": 6590465956085760.0, | |
| "train_loss": 3.374648117792039, | |
| "train_runtime": 765.0166, | |
| "train_samples_per_second": 27.45, | |
| "train_steps_per_second": 2.745 | |
| } | |
| ], | |
| "max_steps": 2100, | |
| "num_train_epochs": 10, | |
| "total_flos": 6590465956085760.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |