MedReason

该数据集是一个大规模、高质量的医疗推理数据集，旨在让大型语言模型（LLM）能够进行可靠且可解释的医疗问题解决。该数据集包含32,682个问题-答案对，每个对都附有详细且逐步的解释。为了生成这个数据集，研究者们利用了一个结构化的医疗知识图谱，将临床问答对转换成逻辑推理链。同时，这些问答对经过了筛选，以确保它们符合一致的 临床逻辑和基于证据的医学原则。该数据集的规模为32,682个问答对，其任务专注于医疗问题解答。

This dataset is a large-scale, high-quality medical reasoning dataset designed to enable large language models (LLMs) to perform reliable and interpretable medical problem-solving. It contains 32,682 question-answer pairs, each accompanied by a detailed, step-by-step explanation. To generate this dataset, researchers utilized a structured medical knowledge graph to convert clinical question-answer pairs into logical reasoning chains. Meanwhile, these question-answer pairs were screened to ensure they align with consistent clinical logic and evidence-based medical principles. The task of this dataset focuses on medical question answering, with a total of 32,682 question-answer pairs included.