ResearchDatasetsRefusalRefusalFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsRefusalRefusal benchmark97.81Alignment Probability55Safety RefusalRefusal0ASR42Safety EvaluationRefusal86.6Refusal Rate36Refusal BehaviorRefusal69.3Sorry Rate22RefusalRefusal (test)64.3BAS (Baseline)6Concept RecoveryREFUSAL0.567Mean MCC6Concept IdentifiabilityRefusal0.9MCC6