Nicholas Renotte: I Spent 6 Weeks using Reinforcement Learning to Beat This GoDot Game

I dove headfirst into training a Stable Baselines 3 agent in GoDotRL over six weeks, sharing all the juicy lessons on hooking it up, reward-shaping hacks, and crafting rock-solid GoDot environments for RL. The full code, logs, and 14 iterations are up on GitHub, plus I’ll happily share those hefty DMG files if you need them.

Along the way I geeked out on a mining-optimization demo, got toasts sent to Claude (we’re cool now), and even prepped for a live talk in Florida. Huge props to Ed Beeching and Brackeys for the inspo—hit me up on LinkedIn, GitHub, Discord, Patreon, or Facebook if you want to nerd out more!

Watch on YouTube