AI Paper Atlas

Research map / Video / Video generation

DiLA: Disentangled Latent Action World Models

Tianqiu Zhang, Muyang Lyu, Yufan Zhang, Fang Fang, Si Wu

ICML 2026 · Video › Video generation · approach: World-model video generation · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add content pathway (disentanglement) to the structure-only latent action model (ablation)DiLA · SSv2+RT-1 mixturerollout LPIPS: 0.344 -> 0.263Improved
add content pathway (ablation)DiLA · SSv2+RT-1 mixtureaction cycle transfer LPIPS: 0.451 -> 0.343Improved
add content pathway (ablation)DiLA · SSv2+RT-1 mixtureMSE after 10k iterations: 0.249 -> 0.216Improved
use temporal-difference bottleneck instead of NSVQ discrete bottleneck (ablation)DiLA · SSv2+RT-1 mixturerollout LPIPS: 0.334 -> 0.263Improved
use temporal-difference bottleneck instead of Gaussian (beta-VAE) bottleneck (ablation)DiLA · SSv2+RT-1 mixturerollout LPIPS: 0.346 -> 0.263Improved
use DiLA bottleneck instead of VQ or Gaussian for latent actions (ablation)DiLA · OOD linear probing (Franka Kitchen)probing MSE: 0.098 / 0.125 -> 0.073Improved
remove IDM and FDM (no latent action learning) (ablation)DiLA · structure-only visualization and rebindingdisentanglement (qualitative)Hurt
add content pathway to structure-only model (ablation)DiLA · RT-1 rolloutSSIM: 0.647 -> 0.774Improved
compare DiLA with AdaWorld (external diffusion model) on SSIMDiLA · SSv2 rolloutSSIM: 0.674 (AdaWorld) vs 0.660 (DiLA)No clear effect
use DiLA instead of AdaWorld as dynamics model for MPCvisual planning with MPPI · VP2 benchmark aggregatesuccess rate (%): 63.50 -> 68.00Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy