PyTorch torch.sparse_csr_tensor 函数
Pytorch torch 参考手册torch.sparse_csr_tensor 是 PyTorch 中用于构造稀疏 CSR（压缩稀疏行）张量的函数。
函数定义
torch.sparse_csr_tensor(crow_indices, col_indices, values, size=None, dtype=None, device=None, requires_grad=False)
使用示例
实例
import torch
# 创建稀疏 CSR 张量
# 行指针
crow_indices = torch.tensor([0, 2, 3])
# 列索引
col_indices = torch.tensor([0, 1, 1])
# 值
values = torch.tensor([1.0, 2.0, 3.0])
# 张量大小
size = (2, 3)
# 构造稀疏 CSR 张量
x = torch.sparse_csr_tensor(crow_indices, col_indices, values, size)
print(x)
print("密集张量:")
print(x.to_dense())
# 创建稀疏 CSR 张量
# 行指针
crow_indices = torch.tensor([0, 2, 3])
# 列索引
col_indices = torch.tensor([0, 1, 1])
# 值
values = torch.tensor([1.0, 2.0, 3.0])
# 张量大小
size = (2, 3)
# 构造稀疏 CSR 张量
x = torch.sparse_csr_tensor(crow_indices, col_indices, values, size)
print(x)
print("密集张量:")
print(x.to_dense())