tofu_Llama-3.1-8B-Instruct_forget05_GradDiff
open-unlearning/tofu_Llama-3.1-8B-Instruct_full unlearned on the TOFU forget05 split with GradDiff, trained with the open-unlearning framework. Used as a weight-unlearning baseline / draft model in the Speculative-Decoding-Unlearning project.
Full training config: .hydra/config.yaml. TOFU evaluation outputs: evals/.
Method hyperparameters
gamma: 1.0
alpha: 5
retain_loss_type: NLL
TOFU summary metrics
| metric | value |
|---|---|
| exact_memorization | 0.4557 |
| extraction_strength | 0.0394 |
| forget_Q_A_PARA_Prob | 0.0003 |
| forget_Q_A_gibberish | 0.4881 |
| forget_quality | 0.0000 |
| forget_truth_ratio | 0.3428 |
| mia_loss | 0.0230 |
| mia_min_k | 0.0235 |
| mia_min_k_plus_plus | 0.0995 |
| mia_zlib | 0.0202 |
| model_utility | 0.5833 |
| privleak | 51.6853 |