ResearchBenchmarksText-to-SQL on Human Evaluation Subset (100 samples)Follow93CPCoTE-SQL84.6886.848991.16Jun 14, 2026Evaluation ResultsMethodMethodLinksCPSSLCEXVESCoTE-SQLBackbone=LLaMA-3.1-8BBackbone=LLaMA-3.1-8B2026.069383867573.61DACBackbone=LLaMA-3.1-8BBackbone=LLaMA-3.1-8B2026.068580797167.59