ResearchTasksPreference EvaluationPreference EvaluationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedUltra-Real benchmarkπbias (rubric-based preference attack)43.1Win Rate2Feb 26, 2026LLM-as-a-Judge comparison setTCRM33.7TCRM Better Rate1Apr 28, 2026PreviousPage 2 of 2Next