Fengchun Qiao

Hi @Aya-S, Thanks for your interests in our work! Please use imageio.imread Best, Fengchun

Hi @finesssss, I think the issue is caused by the version of MetaNN. Please use MetaNN 0.1.5 and try again.

Hi @952942497 , Thanks for your interests in our work! Here is the processed data: https://drive.google.com/open?id=1zE7JNGtLiWuuGdQUrhNyhY4AtY8fk-XB Best, Fengchun

Hi @Neronjust2017, Thanks for pointing out it. The training of WAE should be the same as https://github.com/1Konny/WAE-pytorch, where "retain_graph=True " is necessary. I accidentally deleted it when cleaning the code....

Hi @HaoyouDeng and @txsing , Thanks for your interests in our work! This is for CIFAR10-C. `parser = argparse.ArgumentParser(description='PyTorch WideResNet Training') parser.add_argument('--dataset', default='cifar10', type=str, help='dataset (cifar10 [default] or cifar100)') parser.add_argument('--epochs',...

Hi @tkong1998, This training strategy enables the model to create domain augmentations gradually away from the source. The idea is similar to the curriculum learning. And these domain augmentations are...

Hi @Vidushi1107 , Sorry for the delay. May I ask that have you changed any other part of the code, such as "setting mode.eval() during training"? It seems that the...

Hi @fanhuiyan, I think the problem may be the MNIST data you used. You may check #12 to see whether the discussion helps. Please use the processed MNIST data (train/test.pkl...

Hi @zhuyi3625 , Sorry for the late reply. The config seems fine to me. You mentioned that all the outputs are the same. It seems the model is not trained...

Hi @cherise215 , The results are indeed ridiculously low. The most common reasons for previous problems are either the package version or the second order gradients. And I can't reach...