List your company

Adversarial Attack on AdvBench Original (test)

0CHRFPFT
-211.52538.5Apr 10, 2026

Evaluation results

MethodLinks
FPFTModel=DeepSeek-7B2026.04010
TSBH (Pass@1)Model=DeepSeek-7B2026.0400
TSBH (Pass@5)Model=DeepSeek-7B2026.0400
FPFTModel=Qwen2.5-7B2026.04023
TSBH (Pass@1)Model=Qwen2.5-7B2026.0403
TSBH (Pass@5)Model=Qwen2.5-7B2026.0402
FPFTModel=Llama3.2-3B2026.04038
TSBH (Pass@1)Model=Llama3.2-3B2026.0400
TSBH (Pass@5)Model=Llama3.2-3B2026.0400
H-CoTModel=Qwen2.5-7B2026.0411
AutoRANModel=Qwen2.5-7B2026.0411
Chain-of-LureModel=Qwen2.5-7B2026.0411
CoT HijackingModel=Qwen2.5-7B2026.0411
OriginalModel=Llama3.2-3B2026.0411
H-CoTModel=Llama3.2-3B2026.0411
AutoRANModel=Llama3.2-3B2026.0411
Chain-of-LureModel=Llama3.2-3B2026.0411
CoT HijackingModel=Llama3.2-3B2026.0411
H-CoTModel=DeepSeek-7B2026.0434
AutoRANModel=DeepSeek-7B2026.0445
Chain-of-LureModel=DeepSeek-7B2026.0444
CoT HijackingModel=DeepSeek-7B2026.0444
LRFTModel=Qwen2.5-7B2026.041614
OriginalModel=Qwen2.5-7B2026.04237
OriginalModel=DeepSeek-7B2026.043622
LRFTModel=DeepSeek-7B2026.044954
LRFTModel=Llama3.2-3B2026.045053