mousecpn / l2a-ot Goto Github PK

Thank you for sharing your excellent work! However, the data available is only the pacs and office-home dataset, could you also upload the data for DG-digit or give me some suggestions? I'm afraid that my reproduce will have some errors and cause unfair results. Thanks a lot!

Hi mousecpn,

thank you for the great work. I was wondering where I could read more about the details in implementing the Sinkhorn distance metric. I could not find any algorithm which describes how this is computed, but I would like to understand your code.

hi, what's the version of python, pytorch and torchvision?

For domain generalization, how could you train the model without same classes between only one source domain and target domain, like Market1501 and DukeMTMC-reID ? Could you release the code of person re-identification experiments？

Hello, Could you please release the h5py code? When I use other code to create h5py file, the accuracy is lower than you reported. And I want to use your framework on other datasets. Thank you so mush!

Hello, thank you for sharing your beautiful work. I can't download the data link you provided : "Please download the data from https://drive.google.com/open?id=0B6x7gtvErXgfUU1WcGY5SzdwZVk". It reminds me that I need access. Could you help me? Thanks a lot !

Hello, I read your paper carefully and the idea is great, but I don’t understand part of the source code, which is about loading D.tar and C.tar. What do these two refer to? If I want to train myself data, do you want to change these two? I look forward to your reply, thank you!