Benign Over-refusal on OR-Bench Hard
11String-match Refusal RateBase draft
Evaluation Results
| Method | Links | |
|---|---|---|
| Base draftDraft Model=Llama-3.1-8B2026.06 | 11 | |
| Safety promptDraft Model=Llama-3.1-8B2026.06 | 14.7 | |
| ALIGNBEAMDraft Model=Llama-3.1-8B, K (Beam Count)=32026.06 | 22.3 | |
| Hard prefixDraft Model=Llama-3.1-8B, K (Beam Count)=12026.06 | 23.7 | |
| LlamaGuard prompt-filterDraft Model=Llama-3.1-8B2026.06 | 34.5 | |
| LlamaGuard response-filterDraft Model=Llama-3.1-8B2026.06 | 34.6 |