/BIPIA

A benchmark for evaluating the robustness of LLMs and defenses to indirect prompt injection attacks.

Primary LanguagePythonOtherNOASSERTION