ResearchDatasetsAlpaca + XSTestFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsSafety DetectionAlpaca + XSTest full (val-selected)0.991AUROC42Refusal DetectionAlpaca + XSTest (val)0.988AUROC42