This is the official version of SDEGen model, the more detailed description and how to train the model on your own dataset is coming soon! (I promise to polish it before April)
By the way, a portion of the code is based on ConfGF, please consider checking their work if you're interested in the Molecular Conformation Generation Problem.
By by the way, recently, I have noticed some people confused the molecular conformation generation and molecular generation, while the former refers to learning P(R|G), the latter is related to learning P(G). Please be careful when conducting research! Have your own taste of work, knowing what is good and what is bad. Don't be the man that 'have no views of one's own, repeat what other says'.