How to Transpose a Tensor in PyTorch?

Tensors are the essential data structure in PyTorch that can have N-dimensional data. Sometimes, users may want to find a transpose of 2D or 3D tensors due to various reasons, such as changing data layout from row to column or vice versa. PyTorch provides a “transpose()” method to compute the transpose of any desired matrix by converting columns to rows and rows to columns.

This article will exemplify the method to find/compute the transpose of various tensors in PyTorch.

How to Transpose a Tensor in PyTorch?

To transpose a particular tensor in PyTorch, first, import the PyTorch library. Then, create a desired 2D or 3D tensor. After that, find/compute the transpose of the tensor using the “transpose()” method. Lastly, display transposed tensor.

The basic syntax of “transpose()” method is:

torch.transpose(, 0, 1)

Here, “0” is the first dimension, and “1” is the second dimension to be transposed.

Go through the next provided examples for a better understanding.

Example 1: Find Transpose of 2D Tensor

In the first example, we will create a 2D tensor and find its transpose. Let’s follow the below step-by-step procedure:

Step 1: Import PyTorch Library
First, import the “torch” library to compute the transpose of the tensor:

import torch

Step 2: Create 2D Tensor
Then, create a 2D tensor using the “torch.tensor()” function and print its elements. Here, we are creating the following “Tens1” 2D tensor:

Tens1 = torch.tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

print(Tens1)

The tensor has been created successfully:

Step 3: Find Transpose of Tensor
Now, use the “transpose()” method to find the transpose of the above-created tensor:

Tens_transpose = torch.transpose(Tens1, 0, 1)

Step 4: Display Transposed Tensor
Finally, print the transposed tensor and view its elements:

print("Transpose of Tens1: \n",Tens_transpose)

The below output shows the transpose of the “Tens1” tensor:

Example 2: Find Transpose of 3D Tensor

In the second example, we will create a 3D tensor and find out its transpose. Let’s follow the provided steps:

Step 1: Define 3D Tensor
First, utilize the “torch.tensor()” function to create a 3D tensor and print its elements. Here, we are creating the following “Tens2” 3D tensor:

Tens2 = torch.tensor([[[1, 3],[5, 7]],[[9, 11],[13, 15]]])

print(Tens2)

This has created a 3D tensor as seen below:

Step 2: Find Transpose of Tensor
Then, find the transpose of the above-created 3D tensor using the “transpose()” method:

Tens_transpose = torch.transpose(Tens2, 0, 1)

Step 3: Display Transposed Tensor
Lastly, print the transposed tensor and view its elements:

print("Transpose of Tens2: \n",Tens_transpose)

According to the below output, the transpose of the “Tens2” tensor has been computed:

We have efficiently explained the method to compute a transpose of 2D or 3D tensors in PyTorch.

Note: You can access our Google Colab Notebook at this link.

Conclusion

To transpose a tensor in PyTorch, first, import the “torch” library. Then, create the desired 2D or 3D tensor and view its elements. Next, use the “transpose()” method to find/compute the transpose of the input tensor. Lastly, print the transposed tensor and view its elements. This blog has exemplified the method to find/compute the transpose of different tensors in PyTorch.

How to Transpose a Tensor in PyTorch?