xirui-li/DrAttack

Official implementation of paper: DrAttack: Prompt Decomposition and Reconstruction Makes Powerful LLM Jailbreakers

语言：JavaScript · GitHub 星标：73 · Fork：14

话题：adversarial-attacks · adversarial-machine-learning · attack · jailbreak · llm · llm-attack · llms · safety

在 GitHub 上查看原仓库