ncalib

A modular Python Library for handling Neural Cellular Automata.

Features

• Modular architecture for Neural Cellular Automata
• Easy-to-use API for building and training NCA models
• Extensible framework for custom implementations

Quickstart

Installation

Install ncalib from PyPI using pip:

pip install ncalib

Some examples and loggers are using Weights and Biases. To use them install the required dependencies using:

pip install ncalib[wandb]

Examples

Examples are found in examples. Currently there are following implementations:

• simple_nca.py - Original NCA reconstructing an image (Mordvintsev et al. (2020))
• simple_classifying_mnist.py - Self-classicying MNIST Digits (Randazzo et al. (2020) )
• image_segmentation.py/image_segmentation_hydra.py - NCA that can segment an image (with and without hydra) (Sandler et al. (2020))
• genome_nca.py - NCA that can be seeded with different genomes (Stovold (2023))

Development

This project uses uv for package management and distribution.

Setup Development Environment

# Clone the repository
git clone https://github.com/dwoiwode/ncalib.git
cd ncalib

# Install dependencies with uv
uv sync

# Install extra dependencies
uv sync  --all-extras

Running Tests

Tests are run using pytest:

uv run pytest tests

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Links

• PyPI Package
• GitHub
• Issues