OpenSpiel: A Framework for Reinforcement Learning in Games

TL;DR

OpenSpiel is a comprehensive framework that provides diverse environments and algorithms for research in reinforcement learning and game theory, supporting various game types and analysis tools.

Contribution

It introduces a unified, versatile platform for reinforcement learning research across multiple game types and includes tools for analyzing learning dynamics and evaluation metrics.

Findings

Supports a wide range of game types and environments

Includes tools for analyzing learning dynamics

Facilitates research in reinforcement learning and game theory

Abstract

OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games. OpenSpiel supports n-player (single- and multi- agent) zero-sum, cooperative and general-sum, one-shot and sequential, strictly turn-taking and simultaneous-move, perfect and imperfect information games, as well as traditional multiagent environments such as (partially- and fully- observable) grid worlds and social dilemmas. OpenSpiel also includes tools to analyze learning dynamics and other common evaluation metrics. This document serves both as an overview of the code base and an introduction to the terminology, core concepts, and algorithms across the fields of reinforcement learning, computational game theory, and search.