ResearchDatasetsLIMAFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsLLM Utility EvaluationLIMA original and rewritten variants (fine-tuning)77.6ARC-E Accuracy20Word CountLIMA (test)97.49Surrogate Acc (Initial Constitution)7Instruction FollowingLIMA (low-resource)81.7Win Rate (bn)7