pytorch - How to perform indexing and slicing for sparse tensor

Question

I'm trying to access elements of the sparse coo tensor.

i = [[0,2]]
v = [np.array([1,2,3,4]),[5,6,7,8]]
a = torch.sparse_coo_tensor(i,v,(4,4))
print(a.to_dense())
tensor([[1, 2, 3, 4],
        [0, 0, 0, 0],
        [5, 6, 7, 8],
        [0, 0, 0, 0]])

But I'm unable to acces neither one element, neither the whole row

print(a[0])
RuntimeError: sparse tensors do not have strides
print(a[0,:])
RuntimeError: sparse tensors do not have strides

How to do it correctly?

question from:https://stackoverflow.com/questions/65845579/how-to-perform-indexing-and-slicing-for-sparse-tensor

Answer 1

Waitting for answers