Safety Evaluation on WildJB
33.2Reference RateALIGNBEAM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ALIGNBEAMDraft model=Finance-Llama3-8B, Domain=finance, LBM-DROP=true, alpha=0.5, N=6, K=3, seed=422026.06 | 33.2 | 41.4 | |
| ALIGNBEAMDraft model=Macro average, LBM-DROP=true, alpha=0.5, K=3, seed=422026.06 | 21 | 31.4 | |
| ALIGNBEAMDraft model=Qwen2-7B-Base, Domain=same-vocab ctrl, LBM-DROP=true, alpha=0.5, N=6, K=3, seed=422026.06 | 20.9 | 32.6 | |
| ALIGNBEAMDraft model=Llama-3.1-8B, Domain=general, LBM-DROP=true, alpha=0.5, N=6, K=3, seed=422026.06 | 19.7 | 30.1 | |
| ALIGNBEAMDraft model=MedLlama-3-8B, Domain=medical, LBM-DROP=true, alpha=0.5, N=6, K=3, seed=422026.06 | 16.2 | 27.1 | |
| ALIGNBEAMDraft model=DeepSeek-Math-7B-Instruct, Domain=math, LBM-DROP=true, alpha=0.5, N=20, K=3, seed=422026.06 | 15 | 25.6 |