Vision transformer in tensorflow

Hi,
I am following steps : Image classification , here the model runs in eager mode , How can I make it run graph mode ?