List your company

Jailbreak Defense on JBC

0ASRPreSafe
-3.67221.11445.970.686Mar 18, 2026

Evaluation results

MethodLinks
PreSafeBase Model=Qwen-8B2026.030
PreSafeBase Model=DS-R1-8B2026.031
PreSafeBase Model=DS-R1-14B2026.031
PreSafeBase Model=Qwen-34B2026.031
PreSafeBase Model=DS-R1-7B2026.032
PreSafeBase Model=Skywork-7B2026.033
R2DBase Model=DS-R1-14B2026.033.3
R2DBase Model=Qwen-8B2026.034.7
R2DBase Model=DS-R1-7B2026.035.1
R2DBase Model=Skywork-7B2026.0314.3
R2DBase Model=DS-R1-8B2026.0316.7
Synonym SubstitutionBase Model=DS-R1-7B2026.0327.3
Synonym SubstitutionBase Model=DS-R1-8B2026.0327.3
SafeChainBase Model=DS-R1-7B2026.0327.5
SafeChainBase Model=Skywork-7B2026.0331.3
R2DBase Model=Qwen-34B2026.0332.9
Synonym SubstitutionBase Model=DS-R1-14B2026.0333.7
SafeChainBase Model=DS-R1-14B2026.0336.4
SafeChainBase Model=Qwen-34B2026.0336.8
Synonym SubstitutionBase Model=Qwen-8B2026.0339.4
SafeChainBase Model=Qwen-8B2026.0341
Smooth LLMBase Model=DS-R1-7B2026.0345.5
Smooth LLMBase Model=DS-R1-14B2026.0346.2
Synonym SubstitutionBase Model=Skywork-7B2026.0346.5
Smooth LLMBase Model=Skywork-7B2026.0352
Erase-And-CheckBase Model=DS-R1-7B2026.0352.6
Smooth LLMBase Model=DS-R1-8B2026.0354.2
Erase-And-CheckBase Model=DS-R1-14B2026.0355.8
SafeChainBase Model=DS-R1-8B2026.0356.6
NoneBase Model=DS-R1-14B2026.0361.6
Perplexity FilterBase Model=DS-R1-14B2026.0361.6
Erase-And-CheckBase Model=Qwen-8B2026.0362
Synonym SubstitutionBase Model=Qwen-34B2026.0365.7
NoneBase Model=DS-R1-7B2026.0366.6
Perplexity FilterBase Model=DS-R1-7B2026.0366.7
Remove Non-DictionaryBase Model=DS-R1-14B2026.0367
Erase-And-CheckBase Model=Skywork-7B2026.0367.7
Smooth LLMBase Model=Qwen-8B2026.0367.8
Erase-And-CheckBase Model=DS-R1-8B2026.0369.1
Remove Non-DictionaryBase Model=DS-R1-7B2026.0370.2
Erase-And-CheckBase Model=Qwen-34B2026.0370.6
NoneBase Model=Skywork-7B2026.0371.4
Perplexity FilterBase Model=Skywork-7B2026.0373.6
Remove Non-DictionaryBase Model=DS-R1-8B2026.0377.2
Perplexity FilterBase Model=Qwen-8B2026.0378.4
Remove Non-DictionaryBase Model=Qwen-34B2026.0379.2
NoneBase Model=Qwen-8B2026.0380.4
Remove Non-DictionaryBase Model=Skywork-7B2026.0381.3
NoneBase Model=DS-R1-8B2026.0381.6
Perplexity FilterBase Model=DS-R1-8B2026.0381.6
Remove Non-DictionaryBase Model=Qwen-8B2026.0387
Perplexity FilterBase Model=Qwen-34B2026.0387.3
NoneBase Model=Qwen-34B2026.0388.7
Smooth LLMBase Model=Qwen-34B2026.0391.8