zou-lab/BioMed-R1-Train

该数据集包括两部分，biomed_r1 和 m23k_clean，用于研究医学大型语言模型中的知识和推理分离。biomed_r1 数据集用于强化学习，而 m23k_clean 数据集用于监督微调。数据集的特征包括 id、question、answer、options 和 reasoning 等。README 文件还提供了数据集的划分、大小和下载大小等信息。这些数据集属于 问题回答 任务类别，并标记为 医学。数据集的语言为英语，遵循 MIT 许可。

This dataset includes two parts, biomed_r1 and m23k_clean, used for research on disentangling knowledge and reasoning in medical large language models. The biomed_r1 dataset is for reinforcement learning, while the m23k_clean dataset is for supervised fine-tuning. The features of the datasets include id, question, answer, options, and reasoning. The README file also provides information on dataset splits, sizes, and download sizes. These datasets are categorized under question-answering tasks and tagged with medical. The language of the datasets is English, and they are licensed under MIT.