Flappy Bird, trained

Q-learning and SARSA learn when to flap, then face narrower pipe gaps to test generalization.

Agent replay

Recorded agents

Reduce positions and velocities to a compact tabular state space.

Saved artifacts · click to inspect

Follow the source README for the original runtime, dependencies, data, and configuration.

git clone https://github.com/eforus-overseer/FlappyBird-RL-Agent.git

Source links point to the original public repository. Credit belongs to the project authors and the dependencies credited there.