PromptBench is our unified library for evaluating and understanding large language models.

Please refer to PromptBench for the code and promptbench docs for the detailed docs.

Basic Usage Dynamic Evaluation: DyVal Prompt Attack