Saved in:
Bibliographic Details
Main Authors: Jeon, Jiwon, Cho, Myungsik, Sung, Youngchul
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.11691
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914387658276864
author Jeon, Jiwon
Cho, Myungsik
Sung, Youngchul
author_facet Jeon, Jiwon
Cho, Myungsik
Sung, Youngchul
contents Offline multi-agent reinforcement learning (MARL) with multi-task datasets is challenging due to varying numbers of agents across tasks and the need to generalize to unseen scenarios. Prior works employ transformers with observation tokenization and hierarchical skill learning to address these issues. However, they underutilize the transformer attention mechanism for inter-agent coordination and rely on a single history token, which limits their ability to capture long-horizon temporal dependencies in partially observable MARL settings. In this paper, we propose STAIRS-Former, a transformer architecture augmented with spatial and temporal hierarchies that enables effective attention over critical tokens while capturing long interaction histories. We further introduce token dropout to enhance robustness and generalization across varying agent populations. Extensive experiments on diverse multi-agent benchmarks, including SMAC, SMAC-v2, MPE, and MaMuJoCo, with multi-task datasets demonstrate that STAIRS-Former consistently outperforms prior methods and achieves new state-of-the-art performance.
format Preprint
id arxiv_https___arxiv_org_abs_2603_11691
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle STAIRS-Former: Spatio-Temporal Attention with Interleaved Recursive Structure Transformer for Offline Multi-task Multi-agent Reinforcement Learning
Jeon, Jiwon
Cho, Myungsik
Sung, Youngchul
Artificial Intelligence
Offline multi-agent reinforcement learning (MARL) with multi-task datasets is challenging due to varying numbers of agents across tasks and the need to generalize to unseen scenarios. Prior works employ transformers with observation tokenization and hierarchical skill learning to address these issues. However, they underutilize the transformer attention mechanism for inter-agent coordination and rely on a single history token, which limits their ability to capture long-horizon temporal dependencies in partially observable MARL settings. In this paper, we propose STAIRS-Former, a transformer architecture augmented with spatial and temporal hierarchies that enables effective attention over critical tokens while capturing long interaction histories. We further introduce token dropout to enhance robustness and generalization across varying agent populations. Extensive experiments on diverse multi-agent benchmarks, including SMAC, SMAC-v2, MPE, and MaMuJoCo, with multi-task datasets demonstrate that STAIRS-Former consistently outperforms prior methods and achieves new state-of-the-art performance.
title STAIRS-Former: Spatio-Temporal Attention with Interleaved Recursive Structure Transformer for Offline Multi-task Multi-agent Reinforcement Learning
topic Artificial Intelligence
url https://arxiv.org/abs/2603.11691