Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
REALM‑Benchrealm-bench.org · Low evidenceREALM is a benchmark for evaluating neural surrogates on realistic spatiotemporal multiphysics flows, featuring 11 high-fidelity datasets across reactive flow scenarios. It systematically tests over 12 neural surrogate families, revealing that performance degrades with increasing dimensionality and that inductive biases dominate over parameter count.
Parse Score