ResearchTasksDownstream evaluationFollowBenchmarksDataset NameSOTA MethodSortMost resultsRecently updatedMost papersApplyDataset NameSOTA methodMetricTrendResultsLast UpdatedSEA-HELMQwen3-4B54.1Indonesian18Jun 4, 20269 downstream tasksSurefire-3B62.6Average Accuracy6Mar 4, 2026lm-evaluation-harness ARC-E, BoolQ, HellaSwag, OBQA, SciQDDP0.381ARC-E Accuracy4Jun 2, 2026