ResearchTasksSafety AuditingFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedLlama-3.2-11B-Vision-Instruct (test)RedDiff w/ alignment (L)41.85Detoxify Score5May 11, 2026Gemini flash 1.5RedDiff w/ alignment (M)81.33Detoxify Score5May 11, 2026Full DatasetStable Diffusion3.09Detoxify Identity Attack Score3May 11, 2026