ResearchTasksMask-quality reviewer alignmentMask-quality reviewer alignmentFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedCorrupted masks (n=96) (test)Gem. 3 Pro79.2Grade-Level Accuracy (Prec)6Jun 16, 2026