IBM/eval-assist

EvalAssist is an open-source project that simplifies using large language models as evaluators (LLM-as-a-Judge) of the output of other large language models by supporting users in iteratively refining evaluation criteria in a web-based user experience.