... method for training models to produce deceptive attention masks. ... we show that our manipulated attention-based explanations deceive ...
確定! 回上一頁