Reproduce “Compliance-to-Code: Enhancing Financial Compliance Checking via Code Generation” 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
Siyuan Li, Jian Chen, Rui Yao, Xuming Hu, Peilin Zhou, Weihua Qiu, Simin Zhang, Chucheng Dong and 3 more
The authors introduce Compliance-to-Code, a dataset of Chinese financial regulations, and FinCheck, a pipeline that structures rules, generates Python checks, retrieves data, and reports results. On a benchmark, one-shot reasoning-augmented code generation from gold structures achieves 78.13% Pass@1 with Qwen3-8B, versus 60.58% for direct prompting. Pass@1 measures unit-test success, not real-world compliance accuracy; predicted structures remain weak on difficult cases.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.