Code for Model zoo HCIAE, CoAtt, and RvA

[davidnvq]

Thanks for the great work.
Could you provide us the code for HCIAE, CoAtt, and RvA which leads to the results reported on the paper?

[simpleshinobu]

First, sorry for my late reply, as you can see, I rarely check github. Please contact me with my email and you can get a quicker reply. Thank you for your interest. It is easy to follow the existing repositories from the original authors' encoders and our principles to implement. For convenience, I reproduce the details and update the repository. Furthermore, I also provide some clues to introduce the details of implementations on VisDial-Bert (the state of the art encoder/model), which help us to get ~78% NDCG on Visual Dialog Challenge 2020 test-std set. As the frameworks are totally different, I suggest you can directly modify that repository. Futhermore, I am interested in your recent work on this field and learn a lot, hope for more communication with you in the future.