| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 9564, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0001, | |
| "loss": 2.7307, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.0001, | |
| "loss": 2.395, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0001, | |
| "loss": 2.2933, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.0001, | |
| "loss": 2.2224, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0001, | |
| "loss": 2.1849, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0001, | |
| "loss": 2.1438, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0001, | |
| "loss": 2.0732, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.0001, | |
| "loss": 2.0202, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.0001, | |
| "loss": 2.0071, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9921, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9794, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9669, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9303, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8634, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8675, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8487, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0001, | |
| "loss": 1.851, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0001, | |
| "loss": 1.8532, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0001, | |
| "loss": 1.843, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 9564, | |
| "total_flos": 1.440849458197758e+18, | |
| "train_loss": 2.0546891480896177, | |
| "train_runtime": 43973.736, | |
| "train_samples_per_second": 13.919, | |
| "train_steps_per_second": 0.217 | |
| } | |
| ], | |
| "max_steps": 9564, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.440849458197758e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |