ResearchDatasetsAttaQAttaQFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsDetoxificationAttaQ benchmark0.045Avg Toxicity (Max)32Toxicity EvaluationAttaQ0.04Max Toxicity Score14Toxicity EvaluationAttaQ 1402 prompts (test)0.042Max Toxicity Score14SafetyAttaQ0.81Average Score12Toxicity MitigationATTAQ0.122Average Max Toxicity9