I do my research about developing new machine learning and neural network model for planning chemical syntheses of stereochemical compound. Since I had downloaded the organic reaction database from https://nextmovesoftware.com/blog/2014/02/27/unleashing-over-a-million-reactions-into-the-wild/ and use this research article's model as reference https://www.nature.com/articles/nature25978
but I stuck for training a large amount of chemical reaction (about 1.5 millions chemical reactions). My question is how to train on a large number of chemical reactions?