ResearchBenchmarksNatural Language to Signal Temporal Logic Generation on STL-DivEn Human Evaluation 1.0Follow62.4AccuracyKGST42.6447.7752.958.03May 27, 2025Evaluation ResultsMethodMethodLinksAccuracyKGST2025.0562.4DeepSeek2025.0555GPT-42025.0553GPT-4+Self-Refinestrategy=Self-Refinestrategy=Self-Refine2025.0551.2GPT-3.52025.0548.4DeepSTL2025.0543.4