PyTorch Layers

Layers are modules that perform operations on input data to build neural networks.

Syntax

By using PyTorch’s .Sequential() method to build a neural network, we can specify layers and activation functions in sequence from input to output as shown below:

import torch
import torch.nn as nn

# Define the model for the neural network
model = nn.Sequential(
    # Define layers and activation functions here
    nn.Linear(in_features, out_features),
    nn.ReLU(),
    nn.Linear(in_features, out_features)
)

Example

Below is an example of a basic multi-layer neural network defined using PyTorch’s nn.Sequential() method, where the data is processed through a specified sequence of layers and activation functions:

import torch
import torch.nn as nn

model = nn.Sequential(
    # First layer: Linear transformation from 56 input features to 128 output features
    nn.Linear(56, 128),  # This layer learns to transform input data
    nn.ReLU(),           # Activation function to introduce non-linearity

    # Second layer: Linear transformation from 128 input features to 26 output features
    nn.Linear(128, 26),  # This layer continues to transform the data
    nn.ReLU(),           # Another activation function for non-linearity

    # Third layer: Linear transformation from 26 input features to 1 output feature
    nn.Linear(26, 1)     # Final output layer that produces the model's prediction
)

Copy to clipboard
Copy to clipboard