Skip to yearly menu bar Skip to main content

Workshop: MATH-AI: The 3rd Workshop on Mathematical Reasoning and AI

ARB: Advanced Reasoning Benchmark for Large Language Models

Tom Sawada · Daniel Paleka · Alexander Havrilla · Pranav Tadepalli · Paula Vidas · Alexander Kranias · John Nay · Kshitij Gupta · Aran Komatsuzaki

Keywords: [ LLM ] [ mathematics ] [ physics ] [ benchmark ] [ academic ]


Large Language Models (LLMs) have demonstrated remarkable performance on various quantitative reasoning and knowledge benchmarks. However, many of these benchmarks are losing utility as LLMs get increasingly high scores, despite not yet reaching expert performance in these domains. We introduce ARB, a novel benchmark composed of advanced reasoning problems in multiple fields. ARB presents a more challenging test than prior benchmarks, featuring problems in mathematics, physics, biology, chemistry, and law. As a subset of ARB, we introduce a challenging set of math and physics problems which require advanced symbolic reasoning and domain knowledge. We evaluate recent models such as GPT-4 and Claude on ARB and demonstrate that current models score well below 50% on more demanding tasks. In order to improve both automatic and assisted evaluation capabilities, we introduce a rubric-based evaluation approach, allowing GPT-4 to score its own intermediate reasoning steps. We find promising agreement between annotators and GPT-4 rubric evaluation scores.

Chat is not available.