World Models

Summary

World models are predictive models of environment dynamics used for reasoning, planning, and policy evaluation.

What The Wiki Currently Believes

  • APTAMI treats configurable predictive world models as essential for autonomous agents.
  • LeWorldModel gives a compact end-to-end JEPA world model from pixels for control.
  • Reconstruction or Semantics? shows that latent-space choice matters for robotic diffusion world models and that semantic latents can be more policy-relevant than reconstruction latents.
  • Beyond Language Modeling reports that unified multimodal pretraining can naturally induce world-modeling capabilities.

Evidence

The corpus moves from conceptual architecture to model selection: build predictive latent dynamics, but choose the latent space according to downstream planning relevance rather than visual fidelity alone.

Open Questions

  • How should long-horizon planning be layered on top of compact latent predictors?
  • Are semantic latents sufficient for control tasks that require precise geometry?