PhysicsEval: Inference-Time Techniques to Improve the Reasoning Proficiency of Large Language Models on Physics Problems

TL;DR

This paper evaluates large language models on physics problems, introduces inference-time techniques including multi-agent verification to enhance reasoning, and presents a new benchmark dataset for physics problem-solving.

Contribution

It introduces a novel multi-agent inference framework and a comprehensive physics problem benchmark to improve and assess LLM reasoning in physics.

Findings

Multi-agent frameworks significantly improve performance on difficult problems.

Inference-time techniques enhance LLM reasoning accuracy.

The new PhysicsEval benchmark contains 19,609 diverse physics problems.

Abstract

The discipline of physics stands as a cornerstone of human intellect, driving the evolution of technology and deepening our understanding of the fundamental principles of the cosmos. Contemporary literature includes some works centered on the task of solving physics problems - a crucial domain of natural language reasoning. In this paper, we evaluate the performance of frontier LLMs in solving physics problems, both mathematical and descriptive. We also employ a plethora of inference-time techniques and agentic frameworks to improve the performance of the models. This includes the verification of proposed solutions in a cumulative fashion by other, smaller LLM agents, and we perform a comparative analysis of the performance that the techniques entail. There are significant improvements when the multi-agent framework is applied to problems that the models initially perform poorly on. Furthermore, we introduce a new evaluation benchmark for physics problems, ${\rm P{\small HYSICS}E{\small VAL}}$, consisting of 19,609 problems sourced from various physics textbooks and their corresponding correct solutions scraped from physics forums and educational websites. Our code and data are publicly available at https://github.com/areebuzair/PhysicsEval.