ResearchDatasetsAggregate of evaluation benchmarksFollowBenchmarksTask NameDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyTask NameDataset NameSOTA ResultTrendResultsVisual Reasoning and CalibrationAggregate of 12 evaluation benchmarks Average80.3Accuracy4