How to Use nn.Sigmoid in PyTorch: Syntax and Example

In PyTorch, use nn.Sigmoid() to apply the sigmoid activation function as a layer in your model. Instantiate it and call it on your input tensor to get output values between 0 and 1.

Syntax

The nn.Sigmoid() class creates a sigmoid activation layer. You instantiate it once and then apply it to input tensors to transform values into the range (0, 1).

Usage pattern:

• sigmoid = nn.Sigmoid() - creates the sigmoid layer.
• output = sigmoid(input_tensor) - applies sigmoid to the input tensor.

python

import torch
import torch.nn as nn

sigmoid = nn.Sigmoid()
input_tensor = torch.tensor([-1.0, 0.0, 1.0, 2.0])
output = sigmoid(input_tensor)
print(output)

Output

tensor([0.2689, 0.5000, 0.7311, 0.8808])

Example

This example shows how to use nn.Sigmoid() to convert raw scores into probabilities between 0 and 1. It demonstrates creating the sigmoid layer, applying it to a tensor, and printing the result.

python

import torch
import torch.nn as nn

# Create sigmoid activation layer
sigmoid = nn.Sigmoid()

# Example input tensor with positive and negative values
input_tensor = torch.tensor([-2.0, -1.0, 0.0, 1.0, 2.0])

# Apply sigmoid activation
output = sigmoid(input_tensor)

print('Input:', input_tensor)
print('Output after sigmoid:', output)

Output

Input: tensor([-2., -1.,  0.,  1.,  2.])
Output after sigmoid: tensor([0.1192, 0.2689, 0.5000, 0.7311, 0.8808])

Common Pitfalls

Common mistakes when using nn.Sigmoid() include:

• Trying to use nn.Sigmoid without parentheses, which creates a class reference instead of an instance.
• Applying sigmoid multiple times accidentally, which can distort outputs.
• Using sigmoid on already normalized data, which may not be necessary.

Always instantiate nn.Sigmoid() before calling it on tensors.

python

import torch
import torch.nn as nn

input_tensor = torch.tensor([0.5, 1.0, 1.5])

# Wrong: missing parentheses, sigmoid is a class, not an instance
# sigmoid = nn.Sigmoid
# output = sigmoid(input_tensor)  # This will raise an error

# Correct way:
sigmoid = nn.Sigmoid()
output = sigmoid(input_tensor)
print(output)

Output

tensor([0.6225, 0.7311, 0.8176])

Quick Reference

Usage	Description
nn.Sigmoid()	Creates a sigmoid activation layer
sigmoid(input_tensor)	Applies sigmoid to input tensor
Output range	Values between 0 and 1
Common use	Final layer for binary classification

Key Takeaways

Instantiate nn.Sigmoid() before applying it to tensors.

Sigmoid outputs values between 0 and 1, useful for probabilities.

Avoid calling nn.Sigmoid without parentheses to prevent errors.

Use sigmoid as an activation layer, often in binary classification models.

Applying sigmoid multiple times is usually a mistake.