ResearchTasksPost-training Safety and Utility AlignmentFor AI assistantsMarkdownContent indexPost-training Safety and Utility AlignmentFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMistral-7BCB0Unsafe Rate (%)7Apr 20, 2026