CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models
Published 18 Sept 2026arXiv:2609.18462
Updated 3 h ago · first seen 17 Sept 2026
paper_01M2Q5D3Z9EQQK8GED1C4CTJEQ
Abstract
-cross Abstract: FastWAM-style world action models enable efficient action-only inference, but generalize poorly under visual distribution shifts. Their reconstruction-oriented representations emphasize appearance-specific details, limiting generalization to unseen scenes and objects. Without observation history, the model also lacks temporal evidence for robustly identifying task-relevant state changes and motion in unfamiliar visual conditions. To address these limitations, we present the Causal Semantic World Action Model (CSWAM), which augments FastWAM with a causal semantic expert built on V-JEPA 2.1. V-JEPA provides temporally grounded representations of semantic state changes and motion with less dependence on appearance-specific details. The expert learns their future evolution from a sparse history of current and past observations and shares the history-derived context with both the video and action streams through causal attention. At inference, CSWAM conditions action denoising on the current video state and observed semantic history, retaining efficient action-only inference. We conduct simulation and real-robot experiments to evaluate generalization under distribution shifts. With embodied pretraining, CSWAM raises Randomized success on RoboTwin 2.0 Clean-to-Randomized transfer from 10.16% to 45.18%, a gain of 35.02 percentage points over FastWAM. Across two real-robot tasks and three OOD difficulty levels, CSWAM improves average success over FastWAM by 42.5 percentage points, from 27.5% to 70.0%.
Organizations
Organizations 0
No organization stated. arXiv metadata does not carry affiliations; an organization is linked only when a model card or lab page cites the paper.
Models
Models introduced or described 0
Inbound described_by relations from model cards and documentation.
No model links this paper yet
Datasets
Datasets used 0
No dataset relation recorded.
Benchmarks
Benchmarks used 0
No benchmark relation recorded.
Code
Repositories & frameworks 0
No repository linked.
Timeline
Timeline 4
- Property changedPaperCSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models
CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models: arxiv announce type changed from cross to replace
Arxiv announce typecross→replacearxiv - Property changedPaperCSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models
CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models: published at changed from 2026-09-17T04:00:00+00:00 to 2026-09-18T04:00:00+00:00
Published17 Sept 2026→18 Sept 2026arxiv - Property changedPaperCSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models
CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models: authors changed from ["Chong Ma", "Jian Zhu", "Jianjun Zhang", "Taiyi Su", "Ti… to ["Chong Ma", "Jian Zhu", "Jianjun Zhang", "Taiyi Su", "Ti…
AuthorsChong Ma, Jian Zhu, Jianjun Zhang, Taiyi Su, Tianbin Liu, Yi Xu, Zitai Huang→Chong Ma, Jian Zhu, Jianjun Zhang, Taiyi Su, Tianbin Liu, Weiyi Lu, Yi Xu, Zitai Huangarxiv - New paperPaperCSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models
New paper: CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models
arxiv
Sources
Sources 1
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.