Neural MMO 2.0 is a massively multi-agent environment for reinforcement learning research. The key feature of this new version is a flexible task system that allows users to define a broad range of objectives and reward signals. We challenge researchers to train agents capable of generalizing to tasks, maps, and opponents never seen during training. Neural MMO features procedurally generated maps with 128 agents in the standard setting and support for up to. Version 2.0 is a complete rewrite of its predecessor with three-fold improved performance and compatibility with CleanRL. We release the platform as free and open-source software with comprehensive documentation available at neuralmmo.github.io and an active community Discord. To spark initial research on this new platform, we are concurrently running a competition at NeurIPS 2023.
@article{arxiv.2311.03736,
title = {Neural MMO 2.0: A Massively Multi-task Addition to Massively Multi-agent Learning},
author = {Joseph Suárez and Phillip Isola and Kyoung Whan Choe and David Bloomin and Hao Xiang Li and Nikhil Pinnaparaju and Nishaanth Kanna and Daniel Scott and Ryan Sullivan and Rose S. Shuman and Lucas de Alcântara and Herbie Bradley and Louis Castricato and Kirsty You and Yuhao Jiang and Qimai Li and Jiaxin Chen and Xiaolong Zhu},
journal= {arXiv preprint arXiv:2311.03736},
year = {2023}
}