Skip to content

Documentation overview

AI-Pong evolves neural networks with a genetic algorithm and uses Pong simulations to evaluate them. These docs cover how the project works, how to train and extend it, and the Python components behind its core behavior.

Learn the project

  • Architecture explains population evaluation, neural networks, simulation, and rendering.
  • Coordinate system describes simulation coordinates and their relationship to Arcade rendering.
  • Fitness function explains how candidate players are scored.
  • Web demo covers browser inference and exporting model checkpoints.

Run and extend AI-Pong

  • Training covers training options, timeouts, population size, and mutation.
  • Animation pipeline describes the source-driven animation workflow.
  • Roadmap records current project direction.
  • Release notes document the 1.4.0 CLI, saved-run formats, and migration notes.

Reference

The API reference is generated from the core Python modules with mkdocstrings.

For setup instructions and a short project introduction, see the home page.