gbyuvd/sabrlo-chem-selfies-training
收藏官方服务：
资源简介：
Valid Bioactives and Natural Product SELFIES数据集是从ChemBL34、COCONUTDB和Supernatural3三个数据库中构建的，包含约100万个有效的SELFIES，每个SELFIES的序列长度不超过25。该数据集经过筛选和清洗，用于ChemMiniQ3-SAbRLo模型的训练，并被分割成6个大致相等的部分以便于在有限计算资源下进行训练。
The Valid Bioactives and Natural Product SELFIES dataset consists of approximately 1 million valid SELFIES with a seq_len of 25 or less, built from ChemBL34, COCONUTDB, and Supernatural3. It has been filtered and cleaned for training the ChemMiniQ3-SAbRLo model and split into 6 approximately equal chunks for training under limited compute resources.
提供机构：
gbyuvd
