Conjugated-xTB
收藏资源简介:
Conjugated-xTB数据集是由三星人工智能研究所蒙特利尔创建的,包含290万个π共轭有机分子。这些分子是由181个手选的π共轭分子片段构建而成,并通过允许的原子索引连接。该数据集涵盖了4-8个片段的分子,最多含有100个重原子。数据集的构建未考虑溶解性,但可以通过给构建块添加烷基链来轻易实现。该数据集用于预训练STGG+模型,并支持生成具有分布式性质的π功能分子。
The Conjugated-xTB dataset was created by Samsung AI Center, Montreal, and comprises 2.9 million π-conjugated organic molecules. These molecules are assembled from 181 hand-picked π-conjugated molecular fragments, linked via allowed atomic indices. This dataset encompasses molecules constructed from 4 to 8 fragments, with a maximum of 100 heavy atoms. Solubility was not considered during the dataset's construction, but can be readily achieved by adding alkyl chains to the constituent building blocks. This dataset is utilized for pre-training the STGG+ model and supports the generation of π-functional molecules with distributed properties.

- 1Generating $π$-Functional Molecules Using STGG+ with Active Learning三星人工智能研究所蒙特利尔 · 2025年



