Text-to-SQL Empowered by Large Language Models: A Benchmark Evaluation

Benchmark Model Rank Results
text-to-sql-on-birdDAIL-SQL + GPT-4#4Execution Accuracy % (Test): 57.41Execution Accuracy % (Dev): 54.76
text-to-sql-on-spiderDAIL-SQL + GPT-4 + Self-Consistency#3Execution Accuracy (Test): 86.6Execution Accuracy (Dev): 84.4…