ResearchTasksMulti-token generationMulti-token generationFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedSciQIA2 → SFT40.8Accuracy5Mar 19, 2026HMathAICL60.4Accuracy5Mar 19, 2026GSM8KIA2 → SFT68.8Accuracy5Mar 19, 2026