ResearchDatasetsHarmBench and AdvBenchHarmBench and AdvBenchFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsJailbreak DefenseHarmBench and AdvBench (test)91.2GCG Score44LLM Inference efficiencyHarmBench-Standard and AdvBench2Slowdown8Generative AI Output SafetyHarmBench and AdvBench (test)82.88Safe Rate8