Behaviour Suite for Reinforcement Learning

TL;DR

The paper presents bsuite, a collection of carefully-designed RL experiments and an open-source toolkit to evaluate and analyze reinforcement learning agents, aiming to improve understanding and development of efficient algorithms.

Contribution

It introduces bsuite, a standardized set of RL benchmarks and an evaluation toolkit, to facilitate reproducible research and better understanding of agent behaviors.

Findings

Provides a scalable, shared benchmark suite for RL research

Includes open-source tools for evaluation and analysis

Supports integration with existing RL frameworks

Abstract

This paper introduces the Behaviour Suite for Reinforcement Learning, or bsuite for short. bsuite is a collection of carefully-designed experiments that investigate core capabilities of reinforcement learning (RL) agents with two objectives. First, to collect clear, informative and scalable problems that capture key issues in the design of general and efficient learning algorithms. Second, to study agent behaviour through their performance on these shared benchmarks. To complement this effort, we open source github.com/deepmind/bsuite, which automates evaluation and analysis of any agent on bsuite. This library facilitates reproducible and accessible research on the core issues in RL, and ultimately the design of superior learning algorithms. Our code is Python, and easy to use within existing projects. We include examples with OpenAI Baselines, Dopamine as well as new reference implementations. Going forward, we hope to incorporate more excellent experiments from the research community, and commit to a periodic review of bsuite from a committee of prominent researchers.