ResearchTasksNLP TasksFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast Updated11 NLP tasks Symbol-Tuning (held-out)ICL+FT86.4Accuracy9Feb 26, 2026Consolidated NLP Tasks (eval)Human-designed Augmentation48.44Score (Single Best Aug)9Feb 26, 2026