ResearchBenchmarksText-to-SQL on Spider Robustness Variants (SYN, REALISTIC, DK) (test)Follow76.9Accuracy (SYN)LEAF-SQL69.51671.43373.3575.267May 10, 2026Evaluation ResultsMethodMethodLinksAccuracy (SYN)Accuracy (REALISTIC)Accuracy (DK)Average AccuracyLEAF-SQLModel=GPT-4oModel=GPT-4o2026.0576.982.777.979.2MAC-SQLModel=GPT-4Model=GPT-42026.0575.28173.676.6DAIL-SQLModel=GPT-4Model=GPT-42026.0572.279.571.274.3Zero-shotModel=GPT-4oModel=GPT-4o2026.0569.872.266.569.5