Reproduce “Adaptive Patching Is Harder Than It Looks For Time-Series Forecasting” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Federico Zucchi, Yi Xie, Chao Zhang, Thomas Lampert, Ziyue Li
The authors compare adaptive patching with tuned fixed-patch versions of EntroPE, HDMixer, and TimeMosaic across eight datasets. They select both configurations on validation data, evaluate fresh runs, and analyze resolution allocation under a fixed budget in a shared-loss model. 79.2% of setting-level MSE differences fall within ±2%, with no statistically significant method-wise differences; uniform versions generally cost less.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.