ResearchDatasetsDANGEROUSQAFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsRed-Teaming (Attack Success Rate)DANGEROUSQA0ASR50Safety EvaluationDANGEROUSQA Llama-2 base15.3Chinese Safety Score8Jailbreak AttackDangerousQA1.01Harmful Rate6Safety EvaluationDangerousQA (test)0.012Harmful Rate3