VinVL: Advancing the state of the art for vision-language modelsmicrosoft.com2 points·productceo··0 commentsOpen articleSaveView on HN