ResearchTasksAction Coreference TrackingFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedTangram 5 utterance length Scone-derived (test)GraphTM57.92Accuracy4May 7, 2026Tangram 3 utterance length Scone-derived (test)REINFORCE68.5Accuracy4May 7, 2026