RESEARCH PAPER

Latent Action Pretraining Through World Modeling

Bahey Tharwat; Yara Nasser; Ali Abouzeid; Ian Reid

Classification

View four quadrants
Major category
WAMs
Architecture
Not applicable
Prediction paradigm
Not applicable
Source review status
Verified from primary sources

Category review. LAWM jointly trains latent action chunks and a separate next-frame world model, using the predictive objective to pretrain an executable policy. This is a world-model-based policy-training method, not a standalone general tokenizer/backbone. The world model is explicitly removed for downstream policy inference. Reading evidence

AT A GLANCE

Contribution

A contribution summary has not been added yet.

Abstract

An abstract has not been added yet.

Affiliations

Not listed in the collection.

BibTeX