ResearchTasksReward ModelingFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedPKU beavertails unperturbedOriginal reward68.75Win Rate2Apr 9, 2026Anthropic HH (unperturbed)Original reward63.28Win Rate2Apr 9, 2026StackExchange (val)StackLLaMA (Initial)0REWARD1Feb 26, 2026Reward Modeling 4 sub-tasks——Primary metric0Feb 26, 2026Page 10 of 10PreviousNext