About input of text

[TitaniumOne]

Thanks for your great job! I want to ask why the input is not (image, text) at forward function, such as output = self.model(image, text) . And what is the scheme of matching text logits and image logits?