ResearchDatasetsMMLU, ScienceQA, GSM8K, and HumanEvalMMLU, ScienceQA, GSM8K, and HumanEvalFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsContinual instruction tuningMMLU, ScienceQA, GSM8K, and HumanEval74.6Average Accuracy3