We have witnessed the booming of transformer architectures being successfully applied in various vision applications. In this line of research, we are investigating what are the crucial points in designing an effective Transformer network, and also how to adaptively apply them in different vision tasks.