ResearchDatasetsMT-Bench, HumanEval, and GSM8KMT-Bench, HumanEval, and GSM8KFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsSpeculative DecodingMT-Bench, HumanEval, and GSM8K Mean4.83Mean Acceptance Length (tau)26