An open-source software for evaluating AI algorithms in multiplayer imperfect information games, including Texas Hold'em poker and Dou Dizhu environments.
RoomAI is an open-source platform for artificial intelligence researchers to evaluate AI algorithms, especially reinforcement learning algorithms in multiplayer imperfect information games. It aims to provide common testbeds for AI methods like deep reinforcement learning and planning techniques.
RoomAI includes environments and rule sets for popular imperfect information games like Texas Hold'em, Dou Dizhu, UNO, Werewolf and more. The goal is for AI agents to learn strategies and deal with the challenges of hidden information and bluffing as in the real world.
By providing common frameworks for these games, RoomAI enables rapid prototyping and testing of the latest AI algorithms. Researchers can compare performance of different techniques and measure progress in tackling imperfect information games. Over time as algorithms improve, RoomAI can track and benchmark their capabilities.
With active development and support from the AI research community, RoomAI hopes to grow into a standardized testbed advancing the state-of-the-art in areas like poker AI across academia and industry.
Here are some alternatives to RoomAI:
Suggest an alternative ❐