Red Teaming against BB-3B on OPT-66B ZS
72.3RSRBRT (e+r)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BRT (e+r)Query limit (NQ)=20,0002023.05 | 72.3 | 45.3 | |
| BRT (e)Query limit (NQ)=20,0002023.05 | 70.8 | 46.4 | |
| SLBackbone=OPT-1.3B, Query limit (NQ)=20,0002023.05 | 41.9 | 55.4 | |
| Offensive Top-NQQuery limit (NQ)=20,0002023.05 | 41.5 | 52.2 | |
| SFSBackbone=OPT-1.3B, Query limit (NQ)=20,0002023.05 | 33.4 | 50 | |
| SFSBackbone=Bloom, Query limit (NQ)=20,0002023.05 | 30.5 | 50.1 | |
| BRT (s+r)Query limit (NQ)=20,0002023.05 | 12.5 | 51 | |
| BRT (s)Query limit (NQ)=20,0002023.05 | 11.4 | 44.3 | |
| RandQuery limit (NQ)=20,0002023.05 | 4.2 | 47.3 |