| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Muses-Bench Deepseek max-turns = 10 | Harness-MU | Privacy100 | 26 | 1mo ago | |
| 12 template–attack datasets (test) | Harness-MU | Privacy100 | 8 | 1mo ago | |
| Synthetic D&D dataset (test) | Accuracy92 | 8 | 5mo ago | ||
| TOFU | SUDOLM | Accuracy98.13 | 5 | 5mo ago | |
| Medical | SUDOLM | Accuracy100 | 5 | 5mo ago | |
| Generated access control dataset (test) | Upload Homework100 | 4 | 5mo ago |