Code and data release for FEABench: Evaluating Language Models on Multiphysics Reasoning Ability. [MATH-AI workshop, NeurIPS 2024] - View it on GitHub