ResearchDatasetsBackdoor behaviorsBackdoor behaviorsFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsBackdoor DetectionBackdoor behaviors (Jailbreak, Watching, Refusal, Password, Country, Sentiment) zero-shot86.8AUROC (zero-shot)36
Backdoor DetectionBackdoor behaviors (Jailbreak, Watching, Refusal, Password, Country, Sentiment) zero-shot86.8AUROC (zero-shot)36