BluffJAX: Adversarial Imperfect Information Games in JAX

BluffJAX provides a JAX-based suite of imperfect information games, including Texas Hold'em, Kuhn Poker, and Kemps, optimized for GPU parallelization. The framework achieves simulation throughputs of hundreds of millions of samples per second. This architecture enables high-scale training for reinforcement learning and tree search algorithms via single and multi-GPU scaling.