How to Perform Matrix Multiplication in PyTorch?

PyTorch offers a dynamic way of working with tensors. It enables users to create and manipulate matrices which are two-dimensional tensors. Matrix multiplication is an operation that takes any two matrices as inputs and creates a new third matrix as output. It is an essential operation in machine learning that is utilized in different applications, such as neural network training, linear regression, etc. PyTorch provides various methods or functions to perform matrix multiplication.

This article will illustrate different methods to perform matrix multiplication in PyTorch.

How to Perform Matrix Multiplication in PyTorch?

To multiply matrices in PyTorch, different methods are used:

However, users need to follow the provided steps first to use these methods:

Step 1: Import PyTorch Library

First, import the “torch” library to perform the matrix multiplication:

import torch

Step 2: Create Matrices(Tensors)

Then, create two desired matrices and print their elements. Here, we are creating “M1” and “M2” matrices (tensors) from the list using the “torch.tensor()” function:

M1 = torch.tensor([[3, 1, 7],[2, 5, 9]])

M2 = torch.tensor([[4, 8],[5, 0],[6, 11]])



This has created two matrices as seen below:

Now, we will use different methods to multiply these matrices.

Method 1: Multiply Matrices Using “torch.matmul()” Method

The “torch.matmul()” method can handle both 2D and higher-dimensional tensors. To multiply matrices, use the “torch.matmul()” method and provide desired tensors as a parameter. Here, we are multiplying “M1” and “M2” matrices (tensors) and storing their product in the “prod” variable:

prod = torch.matmul(M1, M2)


In the below output, the product of two matrices can be seen:

Method 2: Multiply Matrices Using “” Method

PyTorch provides a “” method that can only handle 2D tensors and performs a simple matrix multiplication. To multiply two matrices using the “” method, pass the desired matrices as parameter:

prod =, M2)


The below screenshot shows the product of “M1” and “M2” matrices:

Method 3: Multiply Matrices Using “@” Operator

Users can also use the “@” operator to multiply two matrices. It is a short form of the “torch.matmul()” method:

prod = M1 @ M2


According to the below output the matrix multiplication has been performed successfully:

We have efficiently explained the methods to perform the matrix multiplication in PyTorch.

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


To perform the matrix multiplication in PyTorch, first, import the “torch” library and create two desired matrices (tensors). Then, use the “torch.matmul()” or “” method or “@” operator to multiply two matrices and display the results. This article has illustrated different methods to perform matrix multiplication in PyTorch.

About the author

Laiba Younas

I have done bachelors in Computer Science. Being passionate about learning new technologies, I am interested in exploring different programming languages and sharing my experience with the world.