.eye()

The .eye() function in PyTorch is used to create a 2D tensor representing an identity matrix. An identity matrix is a square matrix in which all elements of the principal diagonal are ones, and all other elements are zeros.

This function is commonly used in linear algebra operations and initializing neural network weights.

Syntax

torch.eye(n, m=None, *, out=None, dtype=None, layout=torch.strided, device=None, requires_grad=False)

• n: Specifies the number of rows in the resulting 2-D tensor.
• m (Optional): Specifies the number of columns in the resulting tensor. Defaults to None, in which case the result is a square matrix with m = n.
• out (Optional): A tensor to store the output. If specified, the result is written into this tensor. Defaults to None.
• dtype (Optional): Specifies the desired data type of the resulting tensor. Defaults to the global PyTorch settings if not provided.
• layout (Optional): Defines the desired memory layout of the tensor. Defaults to torch.strided.
• device (Optional): The desired device on which to create the tensor. If not specified, the tensor will be created on the current device.
• requires_grad (Optional): If True, the resulting tensor will be created with gradient computation enabled. Defaults to False.

Example

Here is an example that demonstrates the use of .eye() to create a 3x3 identity matrix:

import torch

# Create a 3x3 identity matrix
identity_matrix = torch.eye(3)

print(identity_matrix)

The output of the above code will be as follows:

tensor([[1., 0., 0.],
        [0., 1., 0.],
        [0., 0., 1.]])

Codebyte Example

Here is an example that demonstrates the use of .eye() to create a 4x4 identity matrix:

us
Visit us
code
Hide code
Code
Output
Hide output
Hide output
Loading...
Run