Hi team,
Thank you for the new sparse api. I wonder whether it is possible to provide a batched version of dgl.sparse.spspmm so that both sparse matrices can have vector values of length K.
In my use case, the sparse matrices are huge, it is impossible to convert one of them into a dense matrix and then use dgl.sparse.bspmm. Could you provide any suggestions? Thanks a lot.
Best,
Simo