Mixture-of-World Models: Scaling Multi-Task Reinforcement Learning with Modular Latent Dynamics

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Boxuan, Zhang, Weipu, Feng, Zhaohan, Xiao, Wei, Sun, Jian, Chen, Jie, Wang, Gang
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917239675944960
author Zhang, Boxuan
Zhang, Weipu
Feng, Zhaohan
Xiao, Wei
Sun, Jian
Chen, Jie
Wang, Gang
author_facet Zhang, Boxuan
Zhang, Weipu
Feng, Zhaohan
Xiao, Wei
Sun, Jian
Chen, Jie
Wang, Gang
contents A fundamental challenge in multi-task reinforcement learning (MTRL) is achieving sample efficiency in visual domains where tasks exhibit substantial heterogeneity in both observations and dynamics. Model-based reinforcement learning offers a promising path to improved sample efficiency through world models, but standard monolithic architectures struggle to capture diverse task dynamics, resulting in poor reconstruction and prediction accuracy. We introduce Mixture-of-World Models (MoW), a scalable architecture that combines modular variational autoencoders for task-adaptive visual compression, a hybrid Transformer-based dynamics model with task-conditioned experts and a shared backbone, and a gradient-based task clustering strategy for efficient parameter allocation. On the Atari 100k benchmark, a single MoW agent trained once on 26 Atari games achieves a mean human-normalized score of 110.4%, competitive with the score of 114.2% achieved by STORM, an ensemble of 26 task-specific models, while using 50% fewer parameters. On Meta-World, MoW achieves a 74.5% average success rate within 300 thousand environment steps, establishing a new state of the art. These results demonstrate that MoW provides a scalable and parameter-efficient foundation for generalist world models.
format Preprint
id arxiv_https___arxiv_org_abs_2602_01270
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Mixture-of-World Models: Scaling Multi-Task Reinforcement Learning with Modular Latent Dynamics
Zhang, Boxuan
Zhang, Weipu
Feng, Zhaohan
Xiao, Wei
Sun, Jian
Chen, Jie
Wang, Gang
Machine Learning
A fundamental challenge in multi-task reinforcement learning (MTRL) is achieving sample efficiency in visual domains where tasks exhibit substantial heterogeneity in both observations and dynamics. Model-based reinforcement learning offers a promising path to improved sample efficiency through world models, but standard monolithic architectures struggle to capture diverse task dynamics, resulting in poor reconstruction and prediction accuracy. We introduce Mixture-of-World Models (MoW), a scalable architecture that combines modular variational autoencoders for task-adaptive visual compression, a hybrid Transformer-based dynamics model with task-conditioned experts and a shared backbone, and a gradient-based task clustering strategy for efficient parameter allocation. On the Atari 100k benchmark, a single MoW agent trained once on 26 Atari games achieves a mean human-normalized score of 110.4%, competitive with the score of 114.2% achieved by STORM, an ensemble of 26 task-specific models, while using 50% fewer parameters. On Meta-World, MoW achieves a 74.5% average success rate within 300 thousand environment steps, establishing a new state of the art. These results demonstrate that MoW provides a scalable and parameter-efficient foundation for generalist world models.
title Mixture-of-World Models: Scaling Multi-Task Reinforcement Learning with Modular Latent Dynamics
topic Machine Learning
url https://arxiv.org/abs/2602.01270