search

HolyBayes / pytorch_ard

Pytorch implementation of Variational Dropout Sparsifies Deep Neural Networks
MIT License
83 stars 16 forks source link

Parallisation #1

Closed fitzgeraldja closed 4 years ago

fitzgeraldja commented 5 years ago

Hi, currently the code isn't parallelisable on multiple devices, as the tensors created in forward passes aren't necessarily assigned to the same device as the input - they need the extra .to(input.device). Otherwise great code, thank you!

HolyBayes commented 4 years ago

Thanks for your comment! Fixed