Fast LeWorldModel

Hugging Face Daily Papers Papers

Summary

Fast-LeWM accelerates visual planning by replacing autoregressive rollout with parallel action-prefix prediction, reducing computational costs and latency accumulation during long-horizon predictions.

Joint-Embedding Predictive Architectures (JEPAs), including recent LeWorldModel (LeWM), have become a promising foundation for reconstruction-free visual world models. For visual planning, however, LeWM evaluates candidate action sequences by repeatedly applying a local one-step latent transition model. This autoregressive rollout makes planning computationally expensive and exposes the predicted trajectory to accumulated latent errors as the horizon grows. We propose Fast LeWorldModel (Fast-LeWM), a fast latent world model that replaces repeated local rollout with action-prefix prediction. Given the current latent and a candidate action sequence, Fast-LeWM encodes its prefixes and predicts the future latents reached after executing those prefixes in parallel. By making action prefixes the basic prediction unit, Fast-LeWM directly models action effects accumulated to different extents over multiple horizons. This prefix-level supervision forces the model to learn how states continuously evolve under different action prefixes, rather than only fitting one-step state transitions. During planning, the predictor can use the last prefix token from the encoded action sequence to evaluate the corresponding future latent without explicitly rolling through each intermediate imagined state. Across multiple tasks, Fast-LeWM improves average success over LeWM while substantially reducing planning time, achieving lower open-loop latent loss whose growth becomes significantly slower as the rollout horizon increases.
Original Article
View Cached Full Text

Cached at: 06/26/26, 02:03 AM

Paper page - Fast LeWorldModel

Source: https://huggingface.co/papers/2606.26217

Abstract

Fast-LeWM accelerates visual planning by replacing autoregressive rollout with parallel action-prefix prediction, reducing computational costs and latency accumulation during long-horizon predictions.

Joint-Embedding Predictive Architectures(JEPAs), including recentLeWorldModel(LeWM), have become a promising foundation for reconstruction-free visual world models. Forvisual planning, however, LeWM evaluates candidate action sequences by repeatedly applying a local one-steplatent transition model. Thisautoregressive rolloutmakes planning computationally expensive and exposes the predicted trajectory to accumulated latent errors as the horizon grows. We propose FastLeWorldModel(Fast-LeWM), a fastlatent world modelthat replaces repeated local rollout withaction-prefix prediction. Given the current latent and a candidate action sequence, Fast-LeWM encodes its prefixes and predicts the future latents reached after executing those prefixes in parallel. By making action prefixes the basic prediction unit, Fast-LeWM directly models action effects accumulated to different extents over multiple horizons. Thisprefix-level supervisionforces the model to learn how states continuously evolve under different action prefixes, rather than only fitting one-step state transitions. During planning, the predictor can use the last prefix token from the encoded action sequence to evaluate the corresponding future latent without explicitly rolling through each intermediate imagined state. Across multiple tasks, Fast-LeWM improves average success over LeWM while substantially reducing planning time, achieving loweropen-loop latent losswhose growth becomes significantly slower as the rollout horizon increases.

View arXiv pageView PDFGitHubAdd to collection

Get this paper in your agent:

hf papers read 2606\.26217

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2606.26217 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2606.26217 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2606.26217 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles

LeWorldModel: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels

Papers with Code Trending

LeWorldModel introduces a stable, end-to-end Joint-Embedding Predictive Architecture that trains directly from pixels with minimal hyperparameters and provable anti-collapse guarantees. It achieves significant speedups in planning compared to foundation models while maintaining competitive performance on robotic manipulation tasks.

Learning Visual Feature-Based World Models via Residual Latent Action

Hugging Face Daily Papers

This paper introduces RLA-WM, a visual feature-based world model that leverages residual latent actions and flow matching to efficiently predict future visual states. The method outperforms existing video-diffusion and feature-based approaches while enabling novel robot learning techniques from offline, actionless demonstration videos.

Looped World Models

Hugging Face Daily Papers

Looped World Models introduce iterative latent state refinement through shared transformer blocks, achieving 100x parameter efficiency while adapting computational depth to prediction complexity.