| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.765486725663717, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.5015040558432561e-05, | |
| "loss": 4.2302, | |
| "step": 1807 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 4.0081586837768555, | |
| "eval_runtime": 33.5458, | |
| "eval_samples_per_second": 3435.628, | |
| "eval_steps_per_second": 13.444, | |
| "step": 1808 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 5.020077553310694e-06, | |
| "loss": 4.0653, | |
| "step": 3614 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 3.9669103622436523, | |
| "eval_runtime": 33.7689, | |
| "eval_samples_per_second": 3412.936, | |
| "eval_steps_per_second": 13.355, | |
| "step": 3616 | |
| } | |
| ], | |
| "max_steps": 5424, | |
| "num_train_epochs": 3, | |
| "total_flos": 3918208057933824.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |