List your company

Benign Query Discrimination on Benign Queries

0.019False Positive Rate (FPR)Guard Few-shot
0.0058040.0948770.183950.273023Feb 26, 2025

Evaluation results

MethodLinks
Guard Few-shotTarget Model=qwen plus2025.020.019
EDDFTarget Model=qwen plus2025.020.0218
Llama3-GuardTarget Model=qwen plus2025.020.083
RegexTarget Model=qwen plus2025.020.0871
EmbeddingTarget Model=qwen plus2025.020.1227
Self-ReminderTarget Model=qwen plus2025.020.1246
Defense PromptTarget Model=qwen plus2025.020.1975
Intention AnalysisTarget Model=qwen plus2025.020.3489