ResearchDatasetsSafe-RLHFFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsStreaming Safety DetectionSafe-RLHF96.43Det@18Safety ModerationSafe RLHF AR92F1 Score8Safety ModerationSafe RLHF EN93F1 Score8Full-response Safety Guardrail ClassificationSafe-RLHF (test)93.2F1 Score7Harmful Query TransformationSafe-RLHF (test)36Effectiveness4Language Model AlignmentSafe RLHF80.7Win Rate (Helpfulness)3