Scalable Evaluation of Multi-Agent Reinforcement Learning with Melting Pot

by   Joel Z. Leibo, et al.

Existing evaluation suites for multi-agent reinforcement learning (MARL) do not assess generalization to novel situations as their primary objective (unlike supervised-learning benchmarks). Our contribution, Melting Pot, is a MARL evaluation suite that fills this gap, and uses reinforcement learning to reduce the human labor required to create novel test scenarios. This works because one agent's behavior constitutes (part of) another agent's environment. To demonstrate scalability, we have created over 80 unique test scenarios covering a broad range of research topics such as social dilemmas, reciprocity, resource sharing, and task partitioning. We apply these test scenarios to standard MARL training algorithms, and demonstrate how Melting Pot reveals weaknesses not apparent from training performance alone.


page 5

page 8

page 9

page 10


Universal Policies to Learn Them All

We explore a collaborative and cooperative multi-agent reinforcement lea...

SMAClite: A Lightweight Environment for Multi-Agent Reinforcement Learning

There is a lack of standard benchmarks for Multi-Agent Reinforcement Lea...

Multi-agent Reinforcement Learning in OpenSpiel: A Reproduction Report

In this report, we present results reproductions for several core algori...

SAT-MARL: Specification Aware Training in Multi-Agent Reinforcement Learning

A characteristic of reinforcement learning is the ability to develop unf...

Generalization of Reinforcement Learners with Working and Episodic Memory

Memory is an important aspect of intelligence and plays a role in many d...

CoachNet: An Adversarial Sampling Approach for Reinforcement Learning

Despite the recent successes of reinforcement learning in games and robo...

Please sign up or login with your details

Forgot password? Click here to reset