from_scipy

iskra.sparse.from_scipy(x: coo_array | csr_array, device: device | str = 'cpu') → SparseTensor[SOURCE]

Converts SciPy sparse COO/CSR arrays into PyTorch sparse tensors.

Parameters:

• x (scipy.sparse.coo_array | scipy.sparse.csr_array) – SciPy sparse array to be converted.

• device (torch.device | str) – Specifies the device on which to construct the tensor. Defaults to “cpu”.

Raises:

ValueError – Throws an exception if the input array is neither COO nor CSR.

Returns:

​ (SparseTensor) – PyTorch tensor with the same data.