PyTorch torch.sparse_csr_tensor Function
Pytorch torch Reference Manual
torch.sparse_csr_tensorIt is a function in PyTorch used to construct sparse CSR (Compressed Sparse Row) tensors.
Function Definition
torch.sparse_csr_tensor(crow_indices, col_indices, values, size=None, dtype=None, device=None, requires_grad=False)
Usage Example
Example
import torch
# Create sparse CSR tensor
# Row pointers
crow_indices = torch.tensor([0, 2, 3])
# Column indices
col_indices = torch.tensor([0, 1, 1])
# Values
values = torch.tensor([1.0, 2.0, 3.0])
# Tensor size
size = (2, 3)
# Construct sparse CSR tensor
x = torch.sparse_csr_tensor(crow_indices, col_indices, values, size)
print(x)
print("Dense tensor:")
print(x.to_dense())
# Create sparse CSR tensor
# Row pointers
crow_indices = torch.tensor([0, 2, 3])
# Column indices
col_indices = torch.tensor([0, 1, 1])
# Values
values = torch.tensor([1.0, 2.0, 3.0])
# Tensor size
size = (2, 3)
# Construct sparse CSR tensor
x = torch.sparse_csr_tensor(crow_indices, col_indices, values, size)
print(x)
print("Dense tensor:")
print(x.to_dense())
Other Extensions