ResearchTasksCoherence evaluationCoherence evaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedHarmBench (test)DirAbl0.22N-gram Repetition Rate16Feb 26, 2026SummEvalMistral55.9Accuracy8Mar 4, 2026