ResearchDatasetsPopulation-levelPopulation-levelFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsModel Repair and RetentionPopulation-level Knowledge & Cognition cohorts (MMLU, TriviaQA, TruthfulQA, ARC-Challenge, GSM8K, IFEval, Math-500, MNLI, HellaSwag)86.9Percent Healed7
Model Repair and RetentionPopulation-level Knowledge & Cognition cohorts (MMLU, TriviaQA, TruthfulQA, ARC-Challenge, GSM8K, IFEval, Math-500, MNLI, HellaSwag)86.9Percent Healed7