Closed xv994 closed 5 months ago
Thank you for the feedback! I have fixed this bug. You can pull the latest code now. This python was written for mdtv1, and we need to change the decoder_layer from 2 to 4 for mdtv2.
Thank you for your reply. The problem has been dealt with. But, unfortunately, I have had another problem again. When I run the 'infer_mdt.py', it reminds me of that. I have no idea how to solve it. I have searched for issues, but there is no useful information.
Please check this one #27 Let me know if you still face this problem
Ohhhh! Maybe I forgot to pip install -e .
. It works now. Thank you for your reply! How nice you are!
Hey! Thank you for your paper and code! They are both great and interesting! But I have some questions, I want to ask for your help.
When I used the MDTv2 weights downloaded from huggingface hub to run the 'infer_mdt.py', I faced this problem. It shows that there are some missing layers and unexpected layers. I just want to know what happens.