ResearchTasksAlignment Reward EvaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedOffsetBias (test)SFT+DPO67.94Reward50Mar 24, 2026HHA (test)SFT+DPO64Harmless Score20Mar 24, 2026