An index of my reinforcement learning projects. Each game lives in its own repository with its own agent, training pipeline, and results.
| Project | Game | Method | Status |
|---|---|---|---|
| flappy-bird-rl | Flappy Bird | DQN (Stable-Baselines3) | Done |
| tic-tac-brainiac | Tic-tac-toe | MENACE-style tabular learning | Done |
| neural-noodle | Snake | PPO, scored on efficiency vs a planner oracle | Done |
| slay-the-spire-gym + slay-the-spire-rl | Slay the Spire | Maskable PPO | In progress |
| donkey-kong-rl | Donkey Kong | PPO + self-imitation learning | Done |
In-progress projects are private until they reach a presentable state, then get added here with links.
Reusable building blocks behind the games above.
| Framework | Purpose | Used by |
|---|---|---|
| matchbox-rl | MENACE-style tabular RL engine for discrete games | tic-tac-brainiac |
MIT.