BackdoorLLM

BackdoorLLM is the first comprehensive benchmark for studying backdoor attacks on Large Language Models (LLMs). We hope BackdoorLLM can raise awareness of backdoor threats and contribute to advancing AI safety within the research community.