Korean Benchmark for Legal Language Understanding (KBL)

• This is an official repository for the KBL dataset from LBox.
• The work will be presented at EMNLP 2024 Findings and the NLLP Workshop.
• An early version of the paper is available here

To Do

• [x] Released the benchmark on the Hugging Face Hub
• [ ] Release task yaml files and push them to the lm-evaluation-harness repository.
• [ ] Publish the evaluation results.
• [ ] Release the Korean statutes and precedents corpus for RAG experiment.
• [ ] Share the data processing script for RAG experiments.

Citation