ResearchTasksContinual Pre-trainingFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedNatural Instructions (val)Anticurriculum2.391Answer Verification7Feb 26, 2026LEGO synthetic (val)Random0Skill 1 Score6Feb 26, 2026