arXiv cs.CL· Tong Liu, Bohan Li, Gongrui Cheng, Zixiong Yu, Yulan He, Volker Tresp·· 14 小时前AI 评分38
LLMs 能解高能理论物理难题，但仍有缺口：QFT-Bench 量子场论评测
LLMs Can Tackle High Energy Theoretical Physics, yet Gaps Remain: Part I, Quantum Field Theory
AI 导读
研究团队推出 QFT-Bench，包含 75 道研究生至研究级量子场论（QFT）推导题，覆盖核心子领域，采用符号验证与 LLM 评判结合评分。在 39 个闭源与开源模型上，最强模型准确率超过 95%，但仍在计算、约定和指令遵循上出错，较弱模型则卡在物理概念与推导层面。对 CoT 的监测还显示模型存在明显的推理不忠实：答案正确，推理过程却不足以支撑结论。
正文 · 原文
Abstract:Beyond math and code, Large language models (LLMs) are beginning to tackle problems that have traditionally required substantial training for theoretical physicists, but it remains unclear how far this capability extends and where it might break. As a first step toward pushing LLMs in high-energy theoretical physics, we introduce QFT-Bench, a benchmark of 75 graduate- to research-level problems requiring extensive reasoning and derivation across core Quantum Field Theory (QFT) subareas. Each question has a well-defined target and closed-form reference answer. Evaluation combines symbolic verification with LLM-based judge. Across 39 proprietary and open-source models, frontier models already demonstrate substantial capabilities on nontrivial QFT calculations, with the strongest models exceeding 95% accuracy. At the same time, clear gaps remain. Even strong models still make errors in calculations, conventions, and instruction following, while weaker models often fail at the level of physical concepts or derivations. We further monitor models' CoTs and show that current models exhibit nontrivial reasoning unfaithfulness, i.e., they could reach correct answers but the reasoning itself does not justify and logically support them. These results provide a quantitative assessment of current LLM capabilities in QFT and a first step toward broader evaluation in high-energy theoretical physics.
| Subjects: | High Energy Physics - Phenomenology (hep-ph); Computation and Language (cs.CL) |
| Cite as: | arXiv:2610.33134 [hep-ph] |
| (or arXiv:2610.33134v1 [hep-ph] for this version) | |
| https://doi.org/10.48550/arXiv.2610.33134 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Tong Liu [view email]
[v1]
Fri, 23 Oct 2026 14:07:03 UTC (931 KB)
来源：arXiv cs.CL · arxiv.org