ResearchBenchmarksText-to-DSL Generation on Spider-DSL (test)Text-to-DSL Generation on Spider-DSL (test)Follow87.3Simple AccuracyREDPARROT52.56461.58270.679.618Mar 7, 2026Evaluation resultsMethodMethodLinksSimple AccuracyModerate AccuracyChallenging AccuracyOverall AccuracyREDPARROTtemplate caching=truetemplate caching=true2026.0387.373.971.377.8ICLtemplate caching=falsetemplate caching=false2026.0353.942.247.447.9