ResearchBenchmarksJailbreaking Attack Detection on Technical Terms Jailbreaking PromptsFollow59Accuracy (MCA)ReGA11.1623.583648.42Jun 2, 2025Evaluation ResultsMethodMethodLinksAccuracy (MCA)Accuracy (MFP)ReGATarget LLM=vicunaTarget LLM=vicuna2025.065939ReGATarget LLM=mistralTarget LLM=mistral2025.065241ReGATarget LLM=llamaTarget LLM=llama2025.065050ReGATarget LLM=qwenTarget LLM=qwen2025.063922ReGATarget LLM=baichuanTarget LLM=baichuan2025.062626ReGATarget LLM=koalaTarget LLM=koala2025.061313