[Tiny Paper] Shortcut World Models: Learning to Leap, Not Step

Pranav Lakshmanan, Paras Chopra

Keywords: World Models, Multi-Horizon Prediction, Discontinuous Dynamics, Model-Based Reinforcement Learning, Learned Simulation

TL;DR: Shortcut World Models predict environment dynamics across multiple horizons in a single forward pass, achieving 33-64x speedup with up to 50% lower error than autoregressive rollout on discontinuous dynamics.

Abstract:

Autoregressive world models chain single-step predictions, requiring N forward
passes for N steps into the future. We introduce Shortcut World Models, trained
to predict environment dynamics across multiple horizons, enabling direct leaping
to any learned step-size in a single pass rather than iteratively stepping through intermediate states. Beyond speed, skipping intermediate predictions also improves
accuracy: errors compound through state discontinuities in autoregressive rollout,
but shortcuts sidestep this accumulation entirely. At inference, adaptive chaining
decomposes arbitrary horizons into learned sub-steps, handling step-sizes beyond
training while maximizing accuracy with minimal sacrifice in speed. On discontinuous particle dynamics, Shortcut World Models achieve 33–64× fewer forward
passes with up to 50% lower error, demonstrating a path toward learned simulators
and model-based planning that are both faster and more accurate.

Submission Number: 92