ResearchTasksMulti-turn RL navigationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedFrozenLake held-out (val)TSR (Beam Search)60.7Success Rate10Feb 26, 2026Sokoban held-out (val)TSR (Beam Search)52.3Success Rate10Feb 26, 2026