ResearchTasksVisual Policy ReasoningFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedClevrPolicy-M N=6TriMPI85.15Accuracy21Apr 21, 2026ClevrPolicy-T (N=6)TriMPI w/ PoRo-DAPO77.8Accuracy14Apr 21, 2026ClevrPolicy-M N=4TriMPI w/ PoRo-DAPO99.4Accuracy7Apr 21, 2026ClevrPolicy-T (N=4)TriMPI w/ PoRo-GRPO99.15Accuracy7Apr 21, 2026