Show HN: A 178K Neural Net that beats Pokémon Roguelike
Category: ai-ml
Tags: reinforcement-learning, pokemon, game-ai, ppo, javascript, agent
Score: 7.3/10 (Innovation: 7, Technical: 9, Documentation: 8, Utility: 5)
A reinforcement learning agent using a tiny 178K-parameter neural net to beat a Pokémon roguelike game, achieved by building a headless simulator from the game's own client code and training with PPO. Its innovation lies in combining game-specific feature engineering with a custom, zero-dependency RL implementation, achieving ~9% win rate versus random play.
Target audience: machine learning engineers, game developers
Repository: https://thiagolira.blot.im/i-got-so-mad-at-poke-rogue-like-that-i-trained-a-rl-agent-to-beat-it-for-me · JavaScript · NOASSERTION
View on Hacker News