ORDerly: Datasets and benchmarks for chemical reaction data

ORDerly provides benchmarks for the most common machine learning tasks with small molecules: Forward prediction, retrosynthesis, and condition prediction!

ORDerly is an open-source Python package for preparing chemical reaction datasets in a reproducible way, and it was recently published at the Neurips AI4Science workshop. It works on top of the increasingly popular Open Reaction Database (ORD). Any questions please do let me know!

Check out our paper here: https://openreview.net/forum?id=R8FQMsECIS&noteId=y1HEbmW7QZ

And our GitHub repo: https://github.com/sustainable-processes/ORDerly