| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| HarmBench | LLaMA Guard 3 | F1 Score98.94 | 100 | 2d ago | |
| XSTEST-RESP | GuardReasoner-Omni 4B | Response Harmfulness F195.48 | 76 | 2d ago | |
| BeaverTails | BeaverDam 7B | F1 Score89.9 | 59 | 2d ago | |
| SafeRLHF | BeaverDam | F1 Score72.1 | 41 | 2d ago | |
| Response Harmfulness Detection Benchmarks (HarmBench, SafeRLHF, BeaverTails, XSTest, WildGuard) | COLAGUARD | Macro Avg F10.8333 | 21 | 5d ago | |
| HarmTextVideo | GuardReasoner-Omni 4B | F1 Score95.25 | 5 | 3mo ago | |
| SPA-VL-Eval | GuardReasoner-Omni 2B | F1 Score74.73 | 5 | 3mo ago |
