yourbench-testing/medical-questions-test

医疗问题测试数据集是一个通过YourBench框架生成的医疗领域基准数据集。该数据集包含三个主要部分：1) 经过标准化处理的原始文档(ingested配置)；2) 被分割成单跳和多跳文本块的分块文本(chunked配置)；3) 使用大型语言模型针对每个文本块生成的独立问答对(single_shot_questions配置)。数据集旨在为医疗领域的自然语言处理任务提供基准测试资源，特别是问答系统开发。生成过程包括文档摄取、文本分块和问题生成三个步骤，确保数据的多样性和专业性。

The Medical Questions Test dataset is a domain-specific benchmark for the medical field generated using the YourBench framework. It consists of three main components: 1) normalized source documents (ingested config), 2) texts split into single-hop and multi-hop chunks (chunked config), and 3) standalone question-answer pairs generated per chunk using LLMs (single_shot_questions config). Designed to support NLP tasks in healthcare, particularly question-answering system development, the dataset was created through a three-step pipeline of document ingestion, text chunking, and question generation to ensure diversity and domain expertise.