ResearchTasksIndividual Label PredictionIndividual Label PredictionFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedHH Helpfulness (consensus-eligible subset)Oracle consensus76.7Per-individual Accuracy4Jul 8, 2026HH Harmlessness consensus-eligible subsetOracle consensus79.8Per-individual Accuracy4Jul 8, 2026