ResearchDatasetsHarmfulQHarmfulQFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsHarmlessness evaluationHarmfulQ (test)100Harmlessness Fraction7Harmfulness EvaluationHarmfulQ0Harmfulness Rate6Safety EvaluationHarmfulQ1.5ASR6JailbreakHarmfulQ18ASR3