List your company

Biomedical Question Answering on PubMedQA (Unauthorized/Authorized/Attack Accuracy)

77Attack AccuracyNo-shield
0.5620.40540.2560.095Oct 16, 2024

Evaluation results

MethodLinks
No-shieldBackbone=LLaMA3-8B2024.1077——
NPLOBackbone=LLaMA3-8B2024.1077——
SerdabBackbone=LLaMA3-8B2024.1072.5——
DarkneTZBackbone=LLaMA3-8B2024.1072.5——
No-shieldBackbone=ChatGLM3-6B2024.1071——
NPLOBackbone=ChatGLM3-6B2024.1070——
No-shieldBackbone=Gemma2-2B2024.1069.5——
NPLOBackbone=Gemma2-2B2024.1069——
SOTERBackbone=LLaMA3-8B2024.1068——
DarkneTZBackbone=ChatGLM3-6B2024.1065.5——
SerdabBackbone=ChatGLM3-6B2024.1063——
SOTERBackbone=ChatGLM3-6B2024.1060.5——
DarkneTZBackbone=Gemma2-2B2024.1060——
No-shieldBackbone=Qwen2-0.5B2024.1058——
NPLOBackbone=Qwen2-0.5B2024.1056.5——
SerdabBackbone=Gemma2-2B2024.1055.5——
SOTERBackbone=Gemma2-2B2024.1055.5——
DarkneTZBackbone=Qwen2-0.5B2024.1051.5——
SerdabBackbone=Qwen2-0.5B2024.1049——
SOTERBackbone=Qwen2-0.5B2024.1047——
CoreGuardBackbone=ChatGLM3-6B2024.1012.5046
DTEBackbone=LLaMA3-8B2024.1012.5——
CoreGuardBackbone=Gemma2-2B2024.1012010.5
ShadowNetBackbone=ChatGLM3-6B2024.1012——
DTEBackbone=ChatGLM3-6B2024.1012——
Black-boxBackbone=ChatGLM3-6B2024.1012——
CoreGuardBackbone=LLaMA3-8B2024.1011029
TLGBackbone=Gemma2-2B2024.1010.5——
Black-boxBackbone=LLaMA3-8B2024.1010.5——
TLGBackbone=ChatGLM3-6B2024.1010——
ShadowNetBackbone=LLaMA3-8B2024.1010——
DTEBackbone=Gemma2-2B2024.109.5——
TLGBackbone=LLaMA3-8B2024.109.5——
ShadowNetBackbone=Gemma2-2B2024.107——
Black-boxBackbone=Gemma2-2B2024.106.5——
CoreGuardBackbone=Qwen2-0.5B2024.106015.5
DTEBackbone=Qwen2-0.5B2024.105.5——
Black-boxBackbone=Qwen2-0.5B2024.105——
ShadowNetBackbone=Qwen2-0.5B2024.104.5——
TLGBackbone=Qwen2-0.5B2024.103.5——