What is LegalEvalHub?

LegalEvalHub is meant to be a platform for sharing and tracking LLM performance on different legal tasks and benchmarks.

Who maintains LegalEvalHub?

LegalEvalHub is maintained by Neel Guha. For questions, comments, or suggestions, please reach out to neelguha@stanford.edu or leave an issue on the GitHub repository.

How do I evaluate my model on LegalEvalHub tasks?

We've created legal-eval-harness, which is a harness for evaluating LLMs on legal tasks. It's very much a work in progress, but it's a good starting point. We'd encourage contributions to the harness in the form of new tasks and model support.

How can I get involved?

We'd love contributions of tasks, leaderboards, and evaluation runs. Please see the contribution guide for more details.