Saved in:
Bibliographic Details
Main Authors: Feng, Kaiyue, Zhao, Yilun, Liu, Yixin, Yang, Tianyu, Zhao, Chen, Sous, John, Cohan, Arman
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2503.21821
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913150465474560
author Feng, Kaiyue
Zhao, Yilun
Liu, Yixin
Yang, Tianyu
Zhao, Chen
Sous, John
Cohan, Arman
author_facet Feng, Kaiyue
Zhao, Yilun
Liu, Yixin
Yang, Tianyu
Zhao, Chen
Sous, John
Cohan, Arman
contents We introduce PHYSICS, a comprehensive benchmark for university-level physics problem solving. It contains 1297 expert-annotated problems covering six core areas: classical mechanics, quantum mechanics, thermodynamics and statistical mechanics, electromagnetism, atomic physics, and optics. Each problem requires advanced physics knowledge and mathematical reasoning. We develop a robust automated evaluation system for precise and reliable validation. Our evaluation of leading foundation models reveals substantial limitations. Even the most advanced model, o3-mini, achieves only 59.9% accuracy, highlighting significant challenges in solving high-level scientific problems. Through comprehensive error analysis, exploration of diverse prompting strategies, and Retrieval-Augmented Generation (RAG)-based knowledge augmentation, we identify key areas for improvement, laying the foundation for future advancements.
format Preprint
id arxiv_https___arxiv_org_abs_2503_21821
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PHYSICS: Benchmarking Foundation Models on University-Level Physics Problem Solving
Feng, Kaiyue
Zhao, Yilun
Liu, Yixin
Yang, Tianyu
Zhao, Chen
Sous, John
Cohan, Arman
Artificial Intelligence
We introduce PHYSICS, a comprehensive benchmark for university-level physics problem solving. It contains 1297 expert-annotated problems covering six core areas: classical mechanics, quantum mechanics, thermodynamics and statistical mechanics, electromagnetism, atomic physics, and optics. Each problem requires advanced physics knowledge and mathematical reasoning. We develop a robust automated evaluation system for precise and reliable validation. Our evaluation of leading foundation models reveals substantial limitations. Even the most advanced model, o3-mini, achieves only 59.9% accuracy, highlighting significant challenges in solving high-level scientific problems. Through comprehensive error analysis, exploration of diverse prompting strategies, and Retrieval-Augmented Generation (RAG)-based knowledge augmentation, we identify key areas for improvement, laying the foundation for future advancements.
title PHYSICS: Benchmarking Foundation Models on University-Level Physics Problem Solving
topic Artificial Intelligence
url https://arxiv.org/abs/2503.21821