some problems with the dataset

Hi, this is a very cool mxnet implementation of JAN and DaNN, however, in this project, you use a dataset which is not easily to get. Would you please share the compare car dataset with me?

And when I try this code on office31 dataset, with the task A->D, I could not re-produce the result shown in the paper (84% in paper, 77 in my try), have you try on office31 dataset and got a nice result?

If so, I am wondering whether you could provide some more information about training? Thank you!

Start by reviewing the repository's dataset and training instructions, focusing on the unavailable comparison-car dataset and the Office31 A→D task mentioned in the issue. Done means documenting how to obtain the dataset and providing enough training information to explain or reproduce the reported 84% result.

Written by the indexing model from the issue text.