ResearchDatasetsVQA, GQA, NLVR, COCOFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsMulti-task Image-Text UnderstandingVQAV2, GQA, NLVR2, COCO Caption66.9VQAv2 Accuracy3