AI Model Architecture

An AI model architecture describes how an AI model is built. It defines which layers the model has and how data flows through them.

Each layer:

1. applies a mathematical rule
2. changes the data a little You can think of layers like filters that data passes through.

Example: Neural Network Architecture

Here is an example of how developers define layers in code. This example is created with pytorch. A very common library used to define networks.

import torch
import torch.nn as nn

class NeuralNetwork(nn.Module):
    def __init__(self):
        super().__init__()

        # Flattens a 28x28 image into a vector of length 784
        self.flatten = nn.Flatten()

        # Stack of linear layers with ReLU activation
        self.linear_relu_stack = nn.Sequential(
            nn.Linear(28 * 28, 512),
            nn.ReLU(),
            nn.Linear(512, 512),
            nn.ReLU(),
            nn.Linear(512, 10)  # 10 output classes
        )

    def forward(self, x):
        x = self.flatten(x)
        logits = self.linear_relu_stack(x)
        return logits

This model has multiple layers stacked together:

1. Flatten layer: Converts the image into a single list of numbers
2. Linear layer: Applies mathematical transformations
3. ReLU layer: Applies a simple rule (keep positive numbers, turn negative to zero)
4. Final layer: Outputs 10 numbers (one for each class)

Learn how this works with classification.