self-ensemble-visual-domain-adapt icon indicating copy to clipboard operation
self-ensemble-visual-domain-adapt copied to clipboard

Published 20 hours ago • verified publisher icon Britefury

Small point about the coding

Open cnyanhao opened this issue 5 years ago • 2 comments

Great work, thank you! Here is my questions. In the script 'network_architectures.py', many 'pool3's are defined but 'pool2's are used instead. Like this

2019-03-06 10 31 08

Maybe it's not very good even though it doesn't matter.

Another question is that I found the 'Self-Ensembling' algorithm had been integrated into the library 'salad' (https://domainadaptation.org/). I'm wondering if this is consistent with your algorithm and if I can use 'salad' directly.

Thank you again for your great job!

cnyanhao avatar Mar 06 '19 02:03 cnyanhao

Woops! Thanks for spotting that!

Thanks for pointing me to Salad. I had seen it a while back and lost the link, so thanks! They seem to get the results that we would expect, although I don't think they use the specific augmentation necessary to crack MNIST -> SVHN. I haven't checked their code yet.

Britefury avatar Mar 11 '19 08:03 Britefury

Hi, I just pushed a new version that fixes the coding issues and is now more compliant with PyTorch v1.0.

Britefury avatar Apr 17 '19 06:04 Britefury