| { |
| "best_metric": 0.5118313494908935, |
| "best_model_checkpoint": "final_models/structroberta_sx2_final/finetune/control_raising_lexical_content_the/checkpoint-400", |
| "epoch": 10.0, |
| "global_step": 570, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 7.02, |
| "eval_accuracy": 0.6695626974105835, |
| "eval_f1": 0.5118313494908935, |
| "eval_loss": 2.921081781387329, |
| "eval_mcc": 0.43375086621448944, |
| "eval_runtime": 43.6835, |
| "eval_samples_per_second": 471.643, |
| "eval_steps_per_second": 58.97, |
| "step": 400 |
| }, |
| { |
| "epoch": 8.77, |
| "learning_rate": 6.140350877192982e-06, |
| "loss": 0.0284, |
| "step": 500 |
| }, |
| { |
| "epoch": 10.0, |
| "step": 570, |
| "total_flos": 6243725097861120.0, |
| "train_loss": 0.024891148131798234, |
| "train_runtime": 354.6918, |
| "train_samples_per_second": 192.167, |
| "train_steps_per_second": 1.607 |
| } |
| ], |
| "max_steps": 570, |
| "num_train_epochs": 10, |
| "total_flos": 6243725097861120.0, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|