Fetching the paper…

CrossFormer++: A Versatile Vision Transformer Hinging on Cross-scale Attention · Around