ResearchTasksSystem-level Correlation with Human JudgmentFor AI assistantsMarkdownContent indexSystem-level Correlation with Human JudgmentFollowBenchmarksDatasetSOTA methodSortMost resultsRecently updatedMost papersApplyDatasetSOTA methodMetricTrendResultsLast updatedMS COCO M2 (system-level)SMURF0.993Pearson r11Feb 26, 2026MS COCO M1 (system-level)BERTScore0.986Pearson r11Feb 26, 2026