| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| ImageNette gas pump Class 7 (test) | SalUn | Forget Accuracy100 | 48 | 2d ago | |
| CIFAR-10 bird, Class 2 (test) | Forgetting Accuracy (Class)100 | 48 | 2d ago | ||
| CIFAR-10 | Accf99.94 | 45 | 2d ago | ||
| TOFU (5%) | OPT-OUT | Forget Quality86.6 | 45 | 3d ago | |
| MNIST | Model Accuracy98.22 | 44 | 3d ago | ||
| TOFU | Forget Quality (FQ)1 | 43 | 3d ago | ||
| CIFAR-100 (test) | Forget Acc98.44 | 43 | 3d ago | ||
| Tiny-ImageNet (train) | FT | Removal Accuracy (Train)100 | 41 | 3d ago | |
| RWKU Llama 3.1 8B (Forget Set) | FB Score85.9 | 39 | 3d ago | ||
| Tiny-ImageNet Swin-T (test) | retrained | Residual Accuracy75.72 | 28 | 3d ago | |
| MUSE-News Llama 2 7B | SimNPO | Privacy Leakage-99.8951 | 27 | 3d ago | |
| MUSE Books | PDU | Privacy Leakage-76.1834 | 25 | 3d ago | |
| UCI Adult | Accuracy89.33 | 24 | 3d ago | ||
| CIFAR-10 | Accuracy76.41 | 24 | 3d ago | ||
| 5% (forget set) | KLM | GF193.398 | 24 | 3d ago | |
| Unlearning Evaluation Dataset Forget, Retain, and Non-member sets | NSPU | SQS83.3 | 24 | 3d ago | |
| CIFAR-10 1.0 (test) | SCRUB | Test Acc95.41 | 24 | 3d ago | |
| TOFU 1.0 (Retain Set) | ROUGE-L100 | 24 | 3d ago | ||
| 32 target sequences (unlearning set) | UL | EL102 | 24 | 3d ago | |
| TOFU (10%) | Retain LLM | Forget Quality (FQ)1 | 23 | 3d ago | |
| ELUDe (retain and world) | RQ51.3 | 22 | 3d ago | ||
| ELUDe (forget) | NPO | FQ89.7 | 22 | 3d ago | |
| CIFAR-10 (train) | Average Gap0 | 22 | 3d ago | ||
| WMDP Cyber (test) | DUET | MMLU60.65 | 21 | 3d ago | |
| CIFAR-100 (test) | Efficacy AUC53.2 | 20 | 3d ago |
