DiLA: Disentangled Latent Action World Models
The paper introduces DiLA, a Disentangled Latent Action world model designed to improve video generation and action abstraction. It addresses the trade-off between action abstraction and generation fidelity by employing content-structure disentanglement. DiLA demonstrates superior performance in various tasks, advancing self-supervised world model learning.
- ▪DiLA resolves the trade-off between action abstraction and generation fidelity through content-structure disentanglement.
- ▪The model distills spatial layouts into a structure pathway while offloading visual details to a content pathway.
- ▪DiLA achieves high-quality video generation, action transfer, and visual planning.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Computer Vision and Pattern Recognition arXiv:2605.15725 (cs) [Submitted on 15 May 2026] Title:DiLA: Disentangled Latent Action World Models Authors:Tianqiu Zhang, Muyang Lyu, Yufan Zhang, Fang Fang, Si Wu View a PDF of the paper titled DiLA: Disentangled Latent Action World Models, by Tianqiu Zhang and 4 other authors View PDF HTML (experimental) Abstract:Latent Action Models (LAMs) enable the learning of world models from unlabeled video by inferring abstract actions between consecutive frames. However, LAMs face a fundamental trade-off between action abstraction and generation fidelity. Existing methods typically circumvent this issue by using two-stage training with pre-trained world models or by limiting predictions to optical flow.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv cs.AI.
Discussion
0 commentsMore from arXiv cs.AI
