ResearchTasksMulti-turn RL Task CompletionFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedFrozenLakeQwen2.5-0.5B (TSR Beam Search)30Success Rate6Feb 26, 2026SokobanQwen2.5-0.5B (TSR Beam Search)38.3Success Rate6Feb 26, 2026