python-pytorchHow can I use Python PyTorch without CUDA?

You can use Python PyTorch without CUDA by setting the device to cpu when creating the Tensors. For example:

import torch

x = torch.rand(5, 3)
print(x.device)

device = torch.device("cpu")
x = torch.rand(5, 3, device=device)
print(x.device)

Output example

cpu
cpu

1. import torch imports the PyTorch library.
2. x = torch.rand(5, 3) creates a tensor of size (5, 3) on the default device, which is usually the GPU.
3. print(x.device) prints the device on which the tensor is located.
4. device = torch.device("cpu") creates a CPU device object.
5. x = torch.rand(5, 3, device=device) creates a tensor of size (5, 3) on the CPU device.
6. print(x.device) prints the device on which the tensor is located.

Helpful links

• https://pytorch.org/docs/stable/tensors.html
• https://pytorch.org/tutorials/beginner/blitz/cifar10_tutorial.html#sphx-glr-beginner-blitz-cifar10-tutorial-py

Edit this code on GitHub