DongkiKim/Mol-LLaMA-Instruct

Mol-LLaMA-Instruct数据集是一个用于微调分子LLM模型的指令数据集，专注于分子结构、化学和生物学特征的关键分子特性，以及LLM在分子推理和解释能力方面的应用。数据集由基于PubChem的注释IUPAC名称及其描述构建而成，包含详细的结构描述、结构到化学特征的关系、结构到生物学特征的关系和全面对话。

The Mol-LLaMA-Instruct dataset is an instruction dataset for fine-tuning molecular LLM models, focusing on key molecular characteristics such as molecular structure, chemical, and biological features, as well as the application of LLM in molecular reasoning and explanability. The dataset is constructed based on annotated IUPAC names and their descriptions from PubChem, including detailed structural descriptions, relationships between structure and chemical features, relationships between structure and biological features, and comprehensive conversations.