python-pytorchHow do I save a PyTorch tensor to a file using Python?

To save a PyTorch tensor to a file using Python, use the following steps:

1. Import the necessary libraries:

import torch
import numpy as np

2. Create a tensor object:

x = torch.tensor([1, 2, 3])

3. Save the tensor to a file:

torch.save(x, 'x_tensor.pt')

4. Verify that the tensor has been saved correctly by loading it back into memory:

x2 = torch.load('x_tensor.pt')
print(x2)

Output example

tensor([1, 2, 3])

5. Save the tensor as a numpy array:

np.save('x_tensor.npy', x.numpy())

6. Verify that the numpy array has been saved correctly by loading it back into memory:

x3 = np.load('x_tensor.npy')
print(x3)

Output example

[1 2 3]

7. Finally, you can also save the tensor to a text file:

torch.save(x, open('x_tensor.txt', 'w'))

Helpful links

• PyTorch Documentation - torch.save
• PyTorch Documentation - torch.load
• Numpy Documentation - np.save
• Numpy Documentation - np.load

Edit this code on GitHub