EU Regulations QA Dataset: Question–Answer Pairs from EU MDR 2017/745 for Legal NLP and Compliance Modeling

This dataset contains 1,039 question–answer pairs derived from Regulation (EU) 2017/745 on medical devices (EU MDR). The dataset was created to support research in legal natural language processing (NLP), regulatory compliance automation, and fine-tuning of large language models (LLMs). The dataset was generated using a retrieval-augmented generation (RAG) pipeline applied to official EU regulatory documents. Each question–answer pair corresponds to specific regulatory provisions and is designed to preserve legal meaning and context. Applications include:- Fine-tuning legal AI models- Question answering systems for regulatory compliance- Legal document analysis- Biomedical regulatory research The dataset is publicly available on Hugging Face:https://huggingface.co/datasets/amal90888/eu-regulations-qa Author: Amal LahchimYear: 2026