MathRobust-LV: Evaluation of Large Language Models' Robustness to Linguistic Variations in Mathematical Reasoning
This addresses the need for reliable deployment of models in educational settings like tutoring and assessment, where instructors rephrase problems, but it is incremental as it builds on existing benchmarking efforts.
The authors tackled the problem of evaluating large language models' robustness to linguistic variations in mathematical reasoning by introducing MathRobust-LV, a test set that rephrases high school-level math problems while keeping difficulty constant, and found that accuracy declines for many models, with drops of 9-11% for smaller ones.
Large language models excel on math benchmarks, but their math reasoning robustness to linguistic variation is underexplored. While recent work increasingly treats high-difficulty competitions like the IMO as the gold standard for evaluating reasoning, we believe in comprehensive benchmarking of high school-level math problems in real educational settings. We introduce MathRobust-LV, a test set and evaluation methodology that mirrors how instructors rephrase problems across assessments while keeping difficulty constant: we change surface details (names, contexts, variables) while preserving numerical structure and answers. In contrast to prior efforts that alter problem content or emphasize IMO-level tasks, we focus on high-school-level dataset problems at the difficulty level where models are currently deployed in educational settings: tutoring and assessment systems. In these applications, instructors rephrase identical concepts in varied ways, making linguistic robustness essential for reliable deployment. Although MATH data benchmarking is often regarded as saturated, our experiment on 34 models reveals that accuracy declines when moving from the baseline to the variants. These drops are severe for smaller models (9-11%) while stronger models also show measurable degradation. Frontier models like GPT-5, Gemini-2.5pro remain comparatively stable. Our results highlight that robustness to linguistic variation is a fundamental challenge, exposing reasoning vulnerabilities in models.