Closed VirajBagal closed 3 years ago
Thank you for sharing the code @calvin-zcx. However, while training from scratch on the zinc250k dataset, after a few batches I get NaN values in the latent vector of the adjacency matrix. Did you preprocess the adjacency matrices in some manner?
Hi, did you find an answer? I would also be interested in it.
Thank you for sharing the code @calvin-zcx. However, while training from scratch on the zinc250k dataset, after a few batches I get NaN values in the latent vector of the adjacency matrix. Did you preprocess the adjacency matrices in some manner?