Tetris Training (Game Boy) — Gymnasium + Stable-Baselines3 + PyBoy

Train reinforcement learning agents to play Tetris for Game Boy using Gymnasium, Stable-Baselines3, and the PyBoy emulator.

This repo is a learning/experiment sandbox focused on reproducible SB3 training, parallel environments, and normalized rewards/observations.

---

Getting Started

Obtaining a ROM

Before training the model, you will need to have legally obtained a ROM of the Tetris game. The file should have a suffix of .gb. Once you have acquired this file you will need to move the file and name it to match the path ./roms/tetris.gb.

Running the Model (WIP)

# 1) install project dependencies
uv sync

# 2) train
uv run -m tetris_training.train

# 3) watch logs
tensorboard --logdir ./tetris_v2/training_data/tboard

References

• Gymnasium - An API standard for reinforcement learning with a diverse collection of reference environments.
• Stable-Baselines3 - a set of reliable implementations of reinforcement learning algorithms in PyTorch.
• PyBoy - A Game Boy emulator written in Python.
• uv - An extremely fast Python package and project manager, written in Rust.

License

MIT. See LICENSE