Saved in:
Bibliographic Details
Main Authors: Cohen, Lior, Wang, Kaixin, Kang, Bingyi, Gadot, Uri, Mannor, Shie
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2502.11537
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915996463267840
author Cohen, Lior
Wang, Kaixin
Kang, Bingyi
Gadot, Uri
Mannor, Shie
author_facet Cohen, Lior
Wang, Kaixin
Kang, Bingyi
Gadot, Uri
Mannor, Shie
contents World models (WMs) represent the frontier of sample-efficient reinforcement learning, but their complexity leaves many promising improvements unrealized due to the significant expertise and effort required to identify and integrate them. Inspired by Rainbow, which showed that individually known improvements to DQN complement each other and can be effectively combined, we take on this challenge and ask whether the same principle applies to world model agents. We introduce Simulus, a modular token-based WM agent that integrates: (1) a flexible tokenization framework supporting arbitrary combinations of observation and action modalities; (2) intrinsic motivation for epistemic uncertainty reduction; (3) prioritized world model replay; and (4) regression-as-classification for reward and return prediction. Simulus achieves state-of-the-art sample efficiency for planning-free WMs across three diverse benchmarks: visual Atari 100K, continuous-control DMC Proprioception 500K, and symbolic Craftax-1M. Notably, intrinsic motivation proves beneficial even under the tight interaction budgets of sample-efficient RL, despite the risk of wasting scarce interactions on task-irrelevant experience. Ablation studies reveal that each component contributes individually, and their combination yields synergistic gains. Our code and model weights are publicly available at https://github.com/leor-c/Simulus.
format Preprint
id arxiv_https___arxiv_org_abs_2502_11537
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Simulus: Combining Improvements in Sample-Efficient World Model Agents
Cohen, Lior
Wang, Kaixin
Kang, Bingyi
Gadot, Uri
Mannor, Shie
Machine Learning
Artificial Intelligence
World models (WMs) represent the frontier of sample-efficient reinforcement learning, but their complexity leaves many promising improvements unrealized due to the significant expertise and effort required to identify and integrate them. Inspired by Rainbow, which showed that individually known improvements to DQN complement each other and can be effectively combined, we take on this challenge and ask whether the same principle applies to world model agents. We introduce Simulus, a modular token-based WM agent that integrates: (1) a flexible tokenization framework supporting arbitrary combinations of observation and action modalities; (2) intrinsic motivation for epistemic uncertainty reduction; (3) prioritized world model replay; and (4) regression-as-classification for reward and return prediction. Simulus achieves state-of-the-art sample efficiency for planning-free WMs across three diverse benchmarks: visual Atari 100K, continuous-control DMC Proprioception 500K, and symbolic Craftax-1M. Notably, intrinsic motivation proves beneficial even under the tight interaction budgets of sample-efficient RL, despite the risk of wasting scarce interactions on task-irrelevant experience. Ablation studies reveal that each component contributes individually, and their combination yields synergistic gains. Our code and model weights are publicly available at https://github.com/leor-c/Simulus.
title Simulus: Combining Improvements in Sample-Efficient World Model Agents
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2502.11537