rlcard

Why we included this project

RLCard is a toolkit for reinforcement learning research in card games, bundling Blackjack, Leduc, Texas Hold'em, Dou Dizhu, Mahjong, and UNO behind Gym-like interfaces. Anyone working on imperfect-information games gets ready-made simulators plus a model zoo of baselines, including DQN, CFR, and rule-based agents, so you can benchmark algorithms without rebuilding game logic. It's also a practical on-ramp for developers prototyping game bots or studying how agents deal with hidden information and multi-player dynamics. Tutorials and a Jupyter notebook walkthrough ease newcomers in, while the tree-search-friendly step-back interface gives advanced users room to write custom CFR-style methods. The pip-installable agent/environment API works in a lecture course or a serious research pipeline.