BluffJAX: Adversarial Imperfect Information Games in JAX

Key points

• We provide canonical implementations of games designed for high simulation throughputs and parallelization on GPU accelerators.1
• We hope that implementing a variety of game mechanics and difficulties will introduce new challenges and foster novel research directions in game-theoretic methods for RL.1
• We benchmark the throughput performance and memory usage of our environments in single and multi-GPU settings, demonstrating scaling of up to hundreds of millions of samples per second, and motivating the usage of BluffJAX over related GPU and CPU-based libraries.1
• We benchmark reinforcement learning, tree search, and game-solving algorithms in JAX in order to provide users with baseline results and facilitate future comparisons.1

Sources (1)

• [1]BluffJAX: Adversarial Imperfect Information Games in JAX
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 6, 03:18 AM

  We introduce BluffJAX: an open-source suite of adversarial imperfect information games in JAX.

  We provide canonical implementations of games designed for high simulation throughputs and parallelization on GPU accelerators.

Extractive summary: sentences quoted from the sources.