// the find
techwithtim/NEAT-Flappy-Bird
An AI that plays flappy bird! Using the NEAT python module.
A single-file demo pairing the NEAT-Python module with Pygame to evolve neural networks that play Flappy Bird, built as a companion to a YouTube tutorial. It's aimed at people learning neuroevolution basics or following along with the video, not at anyone looking for a reusable NEAT framework.
The whole thing lives in one readable flappy_bird.py plus a config-feedforward.txt, so you can see the fitness function, network inputs, and game loop without digging through abstraction layers. It ships a pre-trained best.pickle and a visualize.py for inspecting the evolved network topology, which is a nice touch most tutorial repos skip. Repl.it/Gitpod badges mean you can run it with zero local setup.
No tests, no CI, and the fork count (488 against 789 stars) tells you this is mostly cloned-and-abandoned by people following the tutorial rather than a project anyone builds on. There's no requirements pinning beyond a bare requirements.txt, so pygame/neat-python version drift will break it over time. Last real activity was October 2024 and the README's only substantive content is now sandwiched around a bootcamp ad, which suggests the repo isn't actively maintained as software so much as left up as tutorial marketing collateral.