| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 62.5, | |
| "global_step": 500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 4.647887323943662e-05, | |
| "loss": 6.5392, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 12.5, | |
| "learning_rate": 4.144869215291751e-05, | |
| "loss": 4.3144, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 18.75, | |
| "learning_rate": 3.641851106639839e-05, | |
| "loss": 3.1382, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 25.0, | |
| "learning_rate": 3.1388329979879276e-05, | |
| "loss": 2.4028, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 31.25, | |
| "learning_rate": 2.6358148893360163e-05, | |
| "loss": 1.856, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 37.5, | |
| "learning_rate": 2.1327967806841047e-05, | |
| "loss": 1.4456, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 43.75, | |
| "learning_rate": 1.6297786720321935e-05, | |
| "loss": 1.1454, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 50.0, | |
| "learning_rate": 1.1267605633802817e-05, | |
| "loss": 0.935, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 56.25, | |
| "learning_rate": 6.237424547283703e-06, | |
| "loss": 0.7969, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 62.5, | |
| "learning_rate": 1.2072434607645876e-06, | |
| "loss": 0.721, | |
| "step": 500 | |
| } | |
| ], | |
| "max_steps": 512, | |
| "num_train_epochs": 64, | |
| "total_flos": 14782480903421952, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |