Reproduce “FlexSQL: Flexible Exploration and Execution Make Better Text-to-SQL Agents” 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

Quang Hieu Pham, Yang He, Ping Nie, Canwen Xu, Davood Rafiei, Yuepeng Wang, Xi Ye, Jocelyn Qiaochu Chen

FlexSQL lets text-to-SQL agents inspect database schemas and values, test partial queries, and revisit plans when evidence changes. It generates varied interpretations and implements them in SQL or Python, grouping candidates by execution output. On Spider2-Snow, gpt-oss-120b reaches 65.44% Majority@16, versus 63.80% for DSR-SQL/DeepSeek-R1; this leaderboard comparison does not match models or compute budgets.

Summary by alphaXiv.

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