| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.64, | |
| "global_step": 100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0002, | |
| "loss": 2.8722, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7951, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7847, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7507, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7355, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.0002, | |
| "loss": 2.7085, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002, | |
| "loss": 2.6827, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0002, | |
| "loss": 2.6429, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0002, | |
| "loss": 2.6901, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.0002, | |
| "loss": 2.688, | |
| "step": 100 | |
| } | |
| ], | |
| "max_steps": 156, | |
| "num_train_epochs": 1, | |
| "total_flos": 6.770832332292096e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |