ResearchTasksSafety GuardrailSafety GuardrailFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedPrompt-side and Response-side benchmarks AggregateStage3-DPO87.8F1 Score12Jul 8, 2026Response-side safety benchmarksStage3-DPO87F1 Score12Jul 8, 2026Prompt-side safety benchmarksStage3-DPO88.6F1 Score12Jul 8, 2026