Reproduce “LLMs Can Tackle High Energy Theoretical Physics, yet Gaps Remain: Part I, Quantum Field Theory” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Tong Liu, Bohan Li, Gongrui Cheng, Zixiong Yu, Yulan He, Volker Tresp

The authors introduce QFT-Bench, a benchmark of 75 difficult, well-specified quantum field theory problems, to test models’ answers and displayed derivations. Each final answer is scored correct or incorrect, with symbolic checking and an LLM judge. GPT-6.1-Sol reaches 100.0% average accuracy and Opus-5.5 98.7%, but these closed-form tasks do not test open-ended research or establish derivation faithfulness.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.