List your company

Adversarial Attack on StrongREJECT Hijacked (test)

0CHRFPFT
-4235077Apr 10, 2026

Evaluation results

MethodLinks
FPFTModel=DeepSeek-7B2026.04025
OriginalModel=Llama3.2-3B2026.0414
AutoRANModel=Qwen2.5-7B2026.04415
AutoRANModel=Llama3.2-3B2026.041629
AutoRANModel=DeepSeek-7B2026.041722
CoT HijackingModel=Llama3.2-3B2026.04229
OriginalModel=DeepSeek-7B2026.042430
CoT HijackingModel=DeepSeek-7B2026.042822
CoT HijackingModel=Qwen2.5-7B2026.043124
Chain-of-LureModel=Llama3.2-3B2026.043330
Chain-of-LureModel=DeepSeek-7B2026.044241
H-CoTModel=DeepSeek-7B2026.044532
H-CoTModel=Qwen2.5-7B2026.044626
H-CoTModel=Llama3.2-3B2026.04467
OriginalModel=Qwen2.5-7B2026.04499
Chain-of-LureModel=Qwen2.5-7B2026.045141
LRFTModel=Qwen2.5-7B2026.046262
FPFTModel=Llama3.2-3B2026.048079
LRFTModel=Llama3.2-3B2026.048198
FPFTModel=Qwen2.5-7B2026.048289
LRFTModel=DeepSeek-7B2026.048998
TSBH (Pass@1)Model=Llama3.2-3B2026.049890
TSBH (Pass@5)Model=Llama3.2-3B2026.049896
TSBH (Pass@1)Model=DeepSeek-7B2026.049997
TSBH (Pass@5)Model=DeepSeek-7B2026.049999
TSBH (Pass@1)Model=Qwen2.5-7B2026.0499100
TSBH (Pass@5)Model=Qwen2.5-7B2026.04100100