not supports multimodal(text+image) input!

[mengxinZHANG]

The current code only supports text input and does not support text+images input. Can you provide the code for this part?
(当前的代码只支持输入文本，不支持文本+图片，我没有找到支持文本+图片作为输入的代码！)

[Usernamezhx]

+1

[naoto0804]

+1
As far as I understand, the proposed model does not process the patch images at the current code here.
Layoutgan++ model also does not process the images (the patch encoding part is commented out)

[flt-ime]

Dear @ningyu1991 @sfdc-ospo-bot ,
I noticed that despite some recent commits, this core issue with the implementation still hasn't been addressed.
Since this issue was first raised over a year ago, it would be great to see the code updated to reflect LayoutDETR's major claims about considering "multimodal desires," especially given that your paper was accepted by ECCV.
This is important for the community to fully understand and build upon your work. Any updates on when we might see the correct version? Thanks!