Hello guys. At first sorry for my english, because it isn’t my native language.
I have question and don’t have answer. I have to solve big linear system with a lot of zero values in process of learning with using backward. I think sparse solver is the best decision for this. And my question is how i can solve sparse systems with autograd. Does PyTorch support this function or can i write a wrapper around maybe scipy.sparse.solve? I am junior in python programming and pytorch but i want to solve my problem and i hope the community of pytorch helps me.