MuMOInstruct
收藏资源简介:
MuMOInstruct是一个高质量指令微调数据集,专注于复杂的多属性分子优化任务。该数据集由俄亥俄州立大学提供,包含多个分子属性优化任务,覆盖了药物开发中关键的分子属性,如脂溶性、药物相似性、血脑屏障透过性、致突变性、肠道吸收性和多巴胺受体D2结合亲和力。数据集通过精心设计的分子对,提供了多种属性的同时优化任务,旨在评估大型语言模型在分子优化任务中的性能。
MuMOInstruct is a high-quality instruction-tuning dataset focused on complex multi-attribute molecular optimization tasks. Provided by The Ohio State University, this dataset encompasses multiple molecular property optimization tasks covering key molecular attributes critical to drug development, including lipophilicity, drug-likeness, blood-brain barrier permeability, mutagenicity, intestinal absorption, and dopamine receptor D2 binding affinity. Leveraging well-designed molecular pairs, the dataset provides simultaneous multi-attribute optimization tasks, and is specifically designed to evaluate the performance of large language models (LLMs) on molecular optimization tasks.




