BASF-AI/PubChemSMILESBitextMining

该数据集包含四个配置，分别是canon-desc、canon-title、iso-desc、iso-title。每个配置都包含测试集，其中canon-desc和iso-desc包含描述字段，canon-title和iso-title包含标题字段。所有配置都包含规范化的SMILES或同分异构体的SMILES字段。数据集主要用于化学分子的描述和标题的规范化SMILES表示学习。

The dataset consists of four configurations: canon-desc, canon-title, iso-desc, and iso-title. Each configuration includes a test split, where canon-desc and iso-desc contain a description field, and canon-title and iso-title contain a title field. All configurations include either canonical SMILES or isomeric SMILES fields. The dataset is primarily used for learning the representation of chemical molecule descriptions and titles in normalized SMILES format.