jkkummerfeld/text2sql-data

Aggregates corrected versions of 10 diverse semantic parsing datasets (Academic, ATIS, Geography, Restaurants, Scholar, Spider, WikiSQL, IMDB, Yelp, Advising) with standardized schema, database instances, and variable annotations for improved evaluation methodology. Provides versioned releases with documented data fixes and contributions tracked via pull requests, enabling reproducible benchmarking across text-to-SQL systems while maintaining backward compatibility for published comparisons.

585 stars. No commits in the last 6 months.