List your company

Benign Request Processing on Aggregated Benign Request Datasets

0.1Benign Macro FPRNo defense
-3.75222.24948.2574.251Jun 15, 2026

Evaluation results

MethodLinks
No defenseBase Model=Qwen2.5-72B2026.060.1
ICDBase Model=Qwen2.5-72B2026.060.5
Self Reminder-OPBase Model=Qwen2.5-72B2026.061.1
RAINBase Model=Qwen2.5-72B2026.061.1
DoubtProbeBase Model=Qwen2.5-72B2026.061.9
Self Reminder-HPBase Model=Qwen2.5-72B2026.062.8
SelfDefend-IntentBase Model=Qwen2.5-72B2026.063.5
SelfDefend-DirectBase Model=Qwen2.5-72B2026.064.3
LlamaGuard4Base Model=Qwen2.5-72B2026.065.3
Goal PrioritizationBase Model=Qwen2.5-72B2026.065.4
LlamaGuard2Base Model=Qwen2.5-72B2026.0659.1
LlamaGuard3Base Model=Qwen2.5-72B2026.0680.9
SmoothLLMBase Model=Qwen2.5-72B2026.0696.4