ManimML/docs/source/visualizing_neural_networks.rst

Visualizing Neural Networks with ManimML
========================================

This is a tutorial on how to make neural network architecture visualizations
and animate common algorithms like the forward pass of a neural network. 
Neural networks are a ubiquitous class of machine learning techniques. 
One of the primary usecases for ManimML is for generating animations of neural network architectures.
We have attempted to construct a simple API for defining neural network architectures 
that should feel native to anyone who has used popular deep learning libraries like Pytorch, Tensorflow, and Keras. 
User's can define a sequence of layers and we prove a system for automatically generating various 
animations of concepts like a forward pass. We also allow the user to change the style of rendered 
architectures and algorithm animations. 

For this tutorial we assume that you have already followed the :doc:`Getting Started <getting_started>` tutorial.
This tutorial goes over several simple topics:

1. Generating a simple feed forward neural network diagram
2. Animating the forward pass of a feed forward neural network
3. Generating a diagram of a convolutional neural network
4. Modifying the default style of a neural network

The topics of other tutorials will include:

1. Creating custom neural network layers
2. Creating custom animations of neural networks

=========================================
Visualizing a Feed Forward Neural Network
=========================================

.. manim:: FeedForwardNetworkScene
    from manim_ml.neural_network.neural_network import NeuralNetwork
    from manim_ml.neural_network.layers import FeedForwardLayer
    
    class FeedForwardNetworkScene(Scene):

        def construct(self):
            neural_network = NeuralNetwork([
                FeedForwardLayer(3),
                FeedForwardLayer(5),
                FeedForwardLayer(2),
                FeedForwardLayer(4)
            ])
            self.add(neural_network)