SOURCE-LINKED INTELLIGENCE

One Patch, Three Roles: What Is Actually Coupled in Autoregressive Time-Series Forecasting?

arXiv · AI, language, vision and robotics · article · Sep 20, 2026 · UTC

Patch-based autoregressive time-series forecasting often ties input representation, learned transitions, and recursive execution to one patch length. We ask which of these roles can be adjusted separately. A supporting atomic-encoding study finds greater sensitivity to model width than to atom grouping on the evaluated grid. Our main finding is that a frozen parent's recursive trajectory is easier to fit than the observed future with lightweight parallel exits. Autoregressive Trajectory Distillation (ATD) turns this into selectable ATD-1/2/4/8 execution, with ATD-1 exactly recovering the paren

Read original source ↗ Open in workspace

recordType

paper

region

Global