ResearchDatasetsDynaBenchFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsPolicy Violation DetectionDynaBench (test)86F1 Score12Safety ClassificationDynaBench (test)75.8F1 Score10Safety EvaluationDynaBench Augmented (test)72.19Accuracy7Policy-grounded safety evaluationDynaBench Original73.9Accuracy5