SocialJax: An Evaluation Suite for Multi-agent Reinforcement Learning in Sequential Social Dilemmas

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Guo, Zihao, Shi, Shuqing, Willis, Richard, Tomilin, Tristan, Leibo, Joel Z., Du, Yali
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915870277632000
author Guo, Zihao
Shi, Shuqing
Willis, Richard
Tomilin, Tristan
Leibo, Joel Z.
Du, Yali
author_facet Guo, Zihao
Shi, Shuqing
Willis, Richard
Tomilin, Tristan
Leibo, Joel Z.
Du, Yali
contents Sequential social dilemmas pose a significant challenge in the field of multi-agent reinforcement learning (MARL), requiring environments that accurately reflect the tension between individual and collective interests. Previous benchmarks and environments, such as Melting Pot, provide an evaluation protocol that measures generalization to new social partners in various test scenarios. However, running reinforcement learning algorithms in traditional environments requires substantial computational resources. In this paper, we introduce SocialJax, a suite of sequential social dilemma environments and algorithms implemented in JAX. JAX is a high-performance numerical computing library for Python that enables significant improvements in operational efficiency. Our experiments demonstrate that the SocialJax training pipeline achieves at least 50\texttimes{} speed-up in real-time performance compared to Melting Pot RLlib baselines. Additionally, we validate the effectiveness of baseline algorithms within SocialJax environments. Finally, we use Schelling diagrams to verify the social dilemma properties of these environments, ensuring that they accurately capture the dynamics of social dilemmas.
format Preprint
id arxiv_https___arxiv_org_abs_2503_14576
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SocialJax: An Evaluation Suite for Multi-agent Reinforcement Learning in Sequential Social Dilemmas
Guo, Zihao
Shi, Shuqing
Willis, Richard
Tomilin, Tristan
Leibo, Joel Z.
Du, Yali
Machine Learning
Artificial Intelligence
Sequential social dilemmas pose a significant challenge in the field of multi-agent reinforcement learning (MARL), requiring environments that accurately reflect the tension between individual and collective interests. Previous benchmarks and environments, such as Melting Pot, provide an evaluation protocol that measures generalization to new social partners in various test scenarios. However, running reinforcement learning algorithms in traditional environments requires substantial computational resources. In this paper, we introduce SocialJax, a suite of sequential social dilemma environments and algorithms implemented in JAX. JAX is a high-performance numerical computing library for Python that enables significant improvements in operational efficiency. Our experiments demonstrate that the SocialJax training pipeline achieves at least 50\texttimes{} speed-up in real-time performance compared to Melting Pot RLlib baselines. Additionally, we validate the effectiveness of baseline algorithms within SocialJax environments. Finally, we use Schelling diagrams to verify the social dilemma properties of these environments, ensuring that they accurately capture the dynamics of social dilemmas.
title SocialJax: An Evaluation Suite for Multi-agent Reinforcement Learning in Sequential Social Dilemmas
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2503.14576