ResearchBenchmarksCommonsense Validation and Explanation on ComVEFollow0.88Performance (F+B -> P+B)T5-base0.809280.827640.8460.86436May 4, 2023Evaluation ResultsMethodMethodLinksPerformance (F+B -> P+B)Performance (F+B -> P+I)Simulatability ScorePerformance (F+I -> P+I)TREU ScoreT5-baseBackbone=T5-baseBackbone=T5-base2023.050.880.527-0.3530.949-0.284BART-baseBackbone=BART-baseBackbone=BART-base2023.050.8120.596-0.2160.864-0.164