@rguha @i_wwwish Was thinking of this one, which was also gdb-13, not 14. https://t.co/gGoIt9VFcc
602 followers
252 followers
Some of the perversity of ML: in this paper, the authors managed to improve their model by training it on many duplicates of the original training set, but with the SMILES strings in a non-canonical order! https://t.co/RudtFk0NYC
229 followers
10,744 followers
Randomized SMILES strings improve the quality of molecular generative models | Journal of Cheminformatics | Full Text https://t.co/gPUa1dKkZy