List your company

Mathematical Reasoning on GSM8k (Usage/Attack Accuracy)

55.95Attack AccuracyNo-shield
-1.998813.045628.0943.1344Oct 16, 2024

Evaluation results

MethodLinks
No-shieldBackbone=ChatGLM3-6B2024.1055.95——
NPLOBackbone=ChatGLM3-6B2024.1055.07——
DarkneTZBackbone=ChatGLM3-6B2024.1054.91——
SOTERBackbone=ChatGLM3-6B2024.1054.55——
NPLOBackbone=LLaMA3-8B2024.1053.83——
SerdabBackbone=ChatGLM3-6B2024.1053.67——
No-shieldBackbone=LLaMA3-8B2024.1053.07——
DarkneTZBackbone=LLaMA3-8B2024.1051.31——
SOTERBackbone=LLaMA3-8B2024.1049.75——
SerdabBackbone=LLaMA3-8B2024.1047.79——
No-shieldBackbone=Gemma2-2B2024.1040.94——
NPLOBackbone=Gemma2-2B2024.1040.5——
DarkneTZBackbone=Gemma2-2B2024.1037.07——
SerdabBackbone=Gemma2-2B2024.1035.18——
SOTERBackbone=Gemma2-2B2024.1032.67——
No-shieldBackbone=Qwen2-0.5B2024.1021.53——
NPLOBackbone=Qwen2-0.5B2024.1020.92——
DarkneTZBackbone=Qwen2-0.5B2024.1016.81——
SerdabBackbone=Qwen2-0.5B2024.1014.96——
SOTERBackbone=Qwen2-0.5B2024.1012.5——
ShadowNetBackbone=Gemma2-2B2024.1010.81——
CoreGuardBackbone=LLaMA3-8B2024.106.22033.11
DTEBackbone=LLaMA3-8B2024.106.09——
TLGBackbone=LLaMA3-8B2024.105.61——
TLGBackbone=Gemma2-2B2024.104.58——
DTEBackbone=Gemma2-2B2024.104.56——
ShadowNetBackbone=LLaMA3-8B2024.104.15——
Black-boxBackbone=LLaMA3-8B2024.104.05——
CoreGuardBackbone=Gemma2-2B2024.103.91030.1
TLGBackbone=ChatGLM3-6B2024.102.84——
CoreGuardBackbone=Qwen2-0.5B2024.102.41015.51
DTEBackbone=Qwen2-0.5B2024.102.36——
Black-boxBackbone=Gemma2-2B2024.101.74——
TLGBackbone=Qwen2-0.5B2024.101.43——
ShadowNetBackbone=Qwen2-0.5B2024.101.34——
Black-boxBackbone=Qwen2-0.5B2024.101.29——
CoreGuardBackbone=ChatGLM3-6B2024.101.04037.13
DTEBackbone=ChatGLM3-6B2024.100.93——
ShadowNetBackbone=ChatGLM3-6B2024.100.43——
Black-boxBackbone=ChatGLM3-6B2024.100.23——