search

ozanciga / dirt-t

A DIRT-T Approach to Unsupervised Domain Adaptation for Pytorch
32 stars 5 forks source link

readme

DOI

dirt-t

Pytorch implementation of the paper A DIRT-T Approach to Unsupervised Domain Adaptation. The code here only partially mirrors the original work. It should be possible to use VADA model and a bit of code reuse from the script vada_train.py to be able to perform the recursive iteration described in the paper.

dependencies:

python==3.7
torch==1.0
tqdm==4.31

data:

Go to the official repo, data/ directory and use download_mnist.py and download_svhn.py to get required .mat files. Place them under data/mnist/ and data/svhn/ folders. Then running python vada_train.py should work.