Reproduce “Knowledge-to-SQL: Enhancing SQL Generation with Data Expert LLM” 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
Zijin Hong, Zheng Yuan, Hao Chen, Qinggang Zhang, Feiran Huang, Xiao Huang
Knowledge-to-SQL adds a data-expert model that generates database-specific facts for existing text-to-SQL systems. It learns from human-annotated knowledge and uses database-execution and SQL-related feedback to refine generation. On GPT-4 with BIRD development data, it raises execution accuracy from 33.25 to 37.94 and VES from 35.92 to 42.15; generated knowledge remains below expert annotation, and real-world generalization is unestablished.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.