Reproduce “Get a GRIP, this will be a long TRIP: A Quantifiable Long-Range Framework for Verifying Over-squashing” 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
Ferran Hernandez Caralt, Simon Heilig, Adrián Bazaga, Asja Fischer, Moshe Eliasof, Pietro Liò
The authors introduce TRIP and GRIP, tasks that certify the graph-information range required for prediction. On ECHO topologies, classical baselines reach MSE 1.0 by range 4 on TRIP tasks, while PHDGN and SWAN remain meaningful learners; the corresponding gap is absent on original ECHO tasks. This comparison tests long-range modeling, but does not identify why rankings differ or establish a universally effective architecture.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.