On the FlowBench lid-driven cavity benchmark, vision-transformer foundation models outperform neural operators in data-limited regimes, but all models generalize poorly to out-of-range Reynolds numbers and geometry generalization is not actually tested.
Recent advances and applications of deep learning methods in materials science
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Geometry Matters: Benchmarking Scientific ML Approaches for Flow Prediction around Complex Geometries
On the FlowBench lid-driven cavity benchmark, vision-transformer foundation models outperform neural operators in data-limited regimes, but all models generalize poorly to out-of-range Reynolds numbers and geometry generalization is not actually tested.