AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

HATSolver: Learning Gröbner Bases with Hierarchical Attention Transformers

Mohamed Malhou, Ludovic Perret, Kristin E. Lauter

ICLR 2026 · Architectures and efficiency › Attention and transformers · approach: Transformer neural operators for PDEs

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace flat attention with 2- or 3-level hierarchical attentionbase transformer · Gröbner basis n=6, rho=0.33sequence accuracyImproved
hierarchical attention for faster trainingbase transformer · Gröbner basis n=6, rho=0.33training steps in 52h: <300K -> ~450K stepsImproved
hierarchical attention on larger instancesbase transformer · Gröbner basis n=7 rho=0.25accuracy: 0% -> 9% (HATSolver-2), 8% (HATSolver-3)Improved
add curriculum learning (ablation)HATSolver-3 · Gröbner basis n=13, rho=0.9exact match accuracy: 33.85 -> 61.2Improved
HATSolver-3 vs classical algorithmsSTD-FGLM/Msolve · n=13 over F7, full densitysuccess rate: 60.8 vs 6.5 (STD-FGLM) and 4.6 (Msolve)Improved
hierarchical attention without curriculum (ablation)HATSolver-3 · n=13 over F7exact match accuracy: 14-34% across test densitiesImproved
add curriculum learning to base transformer (ablation)base transformer · Gröbner basis n=7 full densityaccuracy: 41% at n=7 rho=1 (prior limit n=5, rho=0.2)Improved
wider (d=1024) vs deeper base transformer (ablation)base transformer · Gröbner basis n=3 rho=0.5training metricsImproved
replace HAT with Hi-Transformer (ablation)HATSolver-3 · n=13 rho=0.4exact match accuracy: HATSolver success vs never learnedHurt
train on non-prime field F16 vs F7HATSolver · Gröbner basis n=5 density 20%exact match accuracy: 52.45 (F7) vs 43.35 (F16) vs 44.01 (F17)Mixed

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy