List your company

Safety Evaluation on Harmful Evaluation Queries

75.2Final HSSFT
0.21619.68339.1558.617May 23, 2026

Evaluation results

MethodLinks
SFTModel=LLaMA3-8B-Instruct2026.0575.2——
SFTModel=LLaMA3-8B-LAT2026.0574——
SFTModel=LLaMA3-8B-ReFAT2026.0574——
SFTModel=LLaMA2-13B-Chat2026.0533.2——
BaseModel=LLaMA3-8B-Instruct2026.0519.2——
BaseModel=LLaMA3-8B-ReFAT2026.0516.2——
Buffer-and-Reinforce (ours)Model=LLaMA3-8B-Instruct2026.058.488.24.3
BaseModel=LLaMA2-13B-Chat2026.056.4——
Buffer-and-Reinforce (ours)Model=LLaMA2-13B-Chat2026.054.385.71.5
Buffer-and-Reinforce (ours)Model=LLaMA3-8B-ReFAT2026.053.587.32.2
Buffer-and-Reinforce (ours)Model=LLaMA3-8B-LAT2026.053.187.62.1