ResearchDatasetsBlindBlindFollowBenchmarksTaskDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyTaskDatasetSOTA resultTrendResultsHeadline GenerationBlind N ≈ 5000 (test)2.07Mean Rating Improvement (%)4Visual ReasoningBlind (test)90.5Line Intersections Accuracy (Test)3Sector classificationBlind91.4Top-1 Accuracy2