Open In App

How to check if a tensor is contiguous or not in PyTorch

Improve
Improve
Like Article
Like
Save
Share
Report

In this article, we are going to see how to check if a tensor is contiguous or not in PyTorch.

A contiguous tensor could be a tensor whose components are stored in a contiguous order without having any empty space between them. We can check if a tensor is contiguous or not by using the Tensor.is_contiguous() method.

Tensor.is_contiguous() method

This method helps us to identify whether a tensor is contiguous or not. This method returns True if a tensor is contiguous else it will return False. Use the below syntax to understand how to check if a tensor is contiguous or not in PyTorch.

Syntax – Tensor.is_contiguous()

Example 1:

In the following program, we are going to check whether a tensor is contiguous or not.

Python3




# import torch library
import torch
  
# create torch tensors
tens_1 = torch.tensor([1., 2., 3., 4., 5.])
  
# display tensors
print("\n First Tensor - ", tens_1)
  
# check this tensor is contiguous or not
output_1 = tens_1.is_contiguous()
  
# display output
print("\n This tensor is contiguous - ", output_1)


Output:

 

Example 2:

In the following program, we are going to see transpose of a tensor is contiguous or not.

Python3




# import torch library
import torch
  
# define a torch tensor
tens = torch.tensor([[10., 20., 30.],
                     [40., 50., 60.]])
  
# transpose of the above defined tensor
tens_transpose = tens.transpose(0, 1)
  
# display tensors
print("\n Original Tensor \n", tens)
print("\n Transpose of original Tensor \n",
      tens_transpose)
  
# check if a tensor and it's transpose are 
# contiguous or not
Output_1 = tens.is_contiguous()
print("\n Original Tensor is contiguous - ", Output_1)
  
Output_2 = tens_transpose.is_contiguous()
print("\n Transpose of original Tensor is contiguous - ", Output_2)


Output:

 



Last Updated : 21 Apr, 2022
Like Article
Save Article
Previous
Next
Share your thoughts in the comments
Similar Reads