ResearchTasksModel Repair and RetentionFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedPopulation-level Knowledge & Cognition cohorts (MMLU, TriviaQA, TruthfulQA, ARC-Challenge, GSM8K, IFEval, Math-500, MNLI, HellaSwag)FAPM86.9Percent Healed7May 21, 2026
Population-level Knowledge & Cognition cohorts (MMLU, TriviaQA, TruthfulQA, ARC-Challenge, GSM8K, IFEval, Math-500, MNLI, HellaSwag)FAPM86.9Percent Healed7May 21, 2026