Red Teaming against BB-3B on BAD
66.4RSRBRT (e+r)
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| BRT (e+r)Query limit (NQ)=20,0002023.05 | 66.4 | 37.6 | |
| BRT (e)Query limit (NQ)=20,0002023.05 | 65.2 | 39.8 | |
| BRT (s+r)Query limit (NQ)=20,0002023.05 | 57.5 | 40 | |
| Offensive Top-NQQuery limit (NQ)=20,0002023.05 | 57.2 | 40.6 | |
| SLBackbone=OPT-1.3B, Query limit (NQ)=20,0002023.05 | 52.6 | 54.9 | |
| BRT (s)Query limit (NQ)=20,0002023.05 | 50.2 | 40.7 | |
| SFSBackbone=Bloom, Query limit (NQ)=20,0002023.05 | 30.2 | 44.3 | |
| SFSBackbone=OPT-1.3B, Query limit (NQ)=20,0002023.05 | 28.6 | 42.5 | |
| RandQuery limit (NQ)=20,0002023.05 | 25.2 | 42.1 |