Brief

Researchers have developed a new knowledge-aware framework to improve Text-to-SQL models, particularly in low-resource environments. This approach constructs a task-specific knowledge base encompassing schema semantics, business logic, and query patterns. By injecting this knowledge into both training and inference, the framework generates diverse synthetic data and enhances model performance, demonstrating significant improvements across seven benchmarks for both open-source and closed-source large language models. AI

Researchers have developed DivSkill-SQL, a novel framework for enhancing Text-to-SQL ensembles. This method optimizes complementary skills by training new agents on examples that the existing ensemble fails on, thereby increasing the probability of generating at least one correct SQL candidate. The framework demonstrated significant improvements, boosting accuracy by up to 11.1 points on Snowflake and 8.3 points on BigQuery when tested with Opus-4.6 and GPT-5.4 base models on the Spider2-Lite dataset. Notably, these optimized skills showed transferability across different SQL dialects and task formulations, with error analysis indicating a reduction in hallucinations and more reliable complementary skills. AI