ResearchTasksPETPETFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedDifferent-environment out-of-domainBagel + Mixed Training87.7Accuracy14Jun 4, 2026AI2-THOR in-domainBagel + Mixed Training97.8Accuracy14Jun 4, 2026