Nitasurin/Mortal-Policy
🤖 AI AgentNitasurin
A high-performance Riichi Mahjong AI powered by Rust and deep reinforcement learning.
Mortal-Policy is a specialized AI engine designed to tackle the strategic depth of Riichi Mahjong. By utilizing deep reinforcement learning, the model learns optimal discard and call policies that rival human professional play. The project is written in Rust, ensuring memory safety and high-speed execution, which is critical for the intensive calculations required during real-time game state evaluation. It draws architectural inspiration from established Mahjong AI benchmarks like Mortal and Suphx, aiming to push the boundaries of policy-based game agents. The repository provides the necessary infrastructure for training and deploying agents capable of navigating the complex probability trees and tactical nuances inherent in Riichi Mahjong, making it a valuable resource for developers interested in game theory and RL applications.
💡Highlights
- ├─Rust-based high-speed inference
- ├─Deep RL policy-based architecture
- └─Inspired by Mortal and Suphx
🎯For
- ├─AI Researchers
- ├─Game Developers
- └─Reinforcement Learning Engineers