Multi-agent Reinforcement Learning in OpenSpiel: A Reproduction Report

TL;DR

This paper validates the reproducibility of core reinforcement learning algorithms in the OpenSpiel framework, providing detailed documentation and source code to ensure accurate replication of results in game learning research.

Contribution

It offers a comprehensive reproduction study of OpenSpiel algorithms, including detailed hyperparameters and source code for exact replication, which was previously lacking.

Findings

Validated OpenSpiel algorithms against original results

Provided detailed hyperparameters and source code

Ensured reproducibility of game learning experiments

Abstract

In this report, we present results reproductions for several core algorithms implemented in the OpenSpiel framework for learning in games. The primary contribution of this work is a validation of OpenSpiel's re-implemented search and Reinforcement Learning algorithms against the results reported in their respective originating works. Additionally, we provide complete documentation of hyperparameters and source code required to reproduce these experiments easily and exactly.