Sign in to view source links and access this dataset
Description
Griffith University's griffith-bigdata provides preprocessed data for five text-to-SQL benchmarks: BIRD, KaggleDBQA, Spider, Spider2-Lite, and Spider2-Snow. This data supports the AV-SQL framework for decomposing complex natural language queries into SQL. The dataset was last updated on May 11, 2026.
Use Cases
Training text-to-SQL models based on the five included benchmark datasets.
Evaluating model performance on complex, decomposed SQL queries as described in the AV-SQL paper.
Benchmarking natural language understanding systems against standardized database query tasks.
Strengths
Data is preprocessed for five established text-to-SQL benchmarks, potentially reducing setup time.
Associated with a published research paper (AV-SQL) and official GitHub repository, providing context.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count and file formats are unknown, which may limit suitability assessment.
Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
Source
Preprocessed from several text-to-SQL benchmarks (BIRD, KaggleDBQA, Spider, Spider2-Lite, Spider2-Snow).
Collection Method
Preprocessing method is associated with the AV-SQL research framework.
Freshness
Last updated 2026-05-11 05:56:33; freshness should be verified.
License is unknown; terms of use must be verified before application.