| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| MMLU | LLaMA-3.1-8B | Accuracy100 | 30 | 3mo ago | |
| TOFU 5% (forget set) | FQ-13.591 | 25 | 5mo ago | ||
| Virtual-Alpaca | Forget Rate44 | 24 | 3mo ago | ||
| ArXiv unlearning Llama2-7B | DF ROUGE82.5 | 18 | 1mo ago | ||
| MT-Bench | Base Model | Fluency5.62 | 18 | 5mo ago | |
| UNCD-Cyber (Retain) | Base Model | In-Domain Accuracy57.19 | 18 | 5mo ago | |
| UNCD-Cyber (Forget) | Base Model | Accuracy61.96 | 18 | 5mo ago | |
| NEWS | NPO+KLR | Verification Memory (VerMem)22.09 | 16 | 5mo ago | |
| BOOKS | GA+GDR | VerMem0 | 16 | 5mo ago | |
| RWKU | R2F | USR89.3 | 16 | 5mo ago | |
| WMDP | LLMU | Delta Score62.4 | 14 | 2mo ago | |
| TOFU (forget01) | FQ1 | 13 | 1mo ago | ||
| TOFU | DF ROUGE93.2 | 13 | 1mo ago | ||
| TOFU | Aggregated Score61 | 13 | 2mo ago | ||
| TOFU (Forget05) | NSRU | Model Utility0.6863 | 11 | 1mo ago | |
| TOFU Llama2-7B (unlearning) | DF ROUGE90.8 | 10 | 1mo ago | ||
| WMDP Cyber | Forget Rate68.5 | 8 | 3mo ago | ||
| TOFU | LUNE | USR89 | 8 | 5mo ago | |
| KnowUnDo | LUNE | USR91.8 | 8 | 5mo ago | |
| EDU-RELAT | Yao-Neg | USR91.6 | 8 | 5mo ago | |
| TOFU Forget10 | FQ6.57 | 7 | 1mo ago | ||
| TOFU Forget05 (retain) | NSRU | R-MCQA55.8 | 6 | 1mo ago | |
| TOFU Forget05 (Forget) | NPO | F-Cloze5.04 | 6 | 1mo ago | |
| TOFU (Forget 10%) | ICUL | RQ7.456 | 5 | 2mo ago | |
| TOFU (Forget 5%) | ICUL | RQ7.456 | 5 | 2mo ago |