List your company

Jailbreak Attack Defense on AutoRAN

0Reasoning Failure Rate (FFR)ThinkingI
-0.6363.6577.9512.243Aug 6, 2025

Evaluation results

MethodLinks
ThinkingIModel=R1-Qwen-14B2025.08068.2
ReasoningGuardModel=R1-Qwen-14B2025.08070.5
ReasoningGuardModel=R1-Qwen-32B2025.08052.3
SafeKeyModel=R1-Qwen-14B2025.082.370.5
ThinkingIModel=R1-Qwen-32B2025.082.384.1
RealSafe-R1Model=R1-Qwen-14B2025.084.545.5
RealSafe-R1Model=R1-Qwen-32B2025.084.540.9
SmoothLLMModel=R1-Qwen-14B2025.086.868.2
ParaphraseModel=R1-Qwen-32B2025.086.868.2
SmoothLLMModel=R1-Qwen-32B2025.089.168.2
No DefenseModel=R1-Qwen-14B2025.0811.484.1
ParaphraseModel=R1-Qwen-14B2025.0811.465.9
Self-ReminderModel=R1-Qwen-14B2025.0811.470.5
No DefenseModel=R1-Qwen-32B2025.0811.475
SAFEPATH-ZSModel=R1-Qwen-14B2025.0813.670.5
SAFEPATH-ZSModel=R1-Qwen-32B2025.0813.672.7
Self-ReminderModel=R1-Qwen-32B2025.0815.984.1