J1-Bench (Ready Jurist One)
An interactive benchmark that tests AI agents in simulated Chinese legal proceedings, from legal consultation to drafting complaints to adjudicating cases.
Description
J1-Bench is an interactive and comprehensive legal benchmark where LLM agents engage in diverse legal scenarios, completing tasks through interactions with various participants under procedural rules. It features interactive legal environments, an eval dataset, and a protocol for holistic legal agent evaluation.
Access the Dataset
https://github.com/FudanDISC/J1BenchJusticeBench lists this dataset for research and development purposes only. Listing it does not constitute an endorsement, recommendation, or approval by the Stanford Legal Design Lab. See the Terms of Use.