ResearchBenchmarksMulti-hop Tool-use on ToolHop unseen (test)Follow43.1AccuracyRimRule34.7836.9439.141.26Dec 31, 2025Evaluation ResultsMethodMethodLinksAccuracyRimRuleAdaptation Method=Recu...Adaptation Method=Recursive Induction of Symbolic Rules2025.1243.1Few-shotAdaptation Method=Few-...Adaptation Method=Few-shot prompting2025.1237.9SEEAdaptation Method=Self...Adaptation Method=Self-Evolutionary Evaluation2025.1235.9Zero-shotAdaptation Method=Zero...Adaptation Method=Zero-shot prompting2025.1235.1