Dmitry Ulyanov

Hey @michaelhuang74 , it is not possible with this code, I cut it for simplicity to be true. Did you find the cases when these weights are important?

Hello, added this feature with https://github.com/DmitryUlyanov/texture_nets/commit/7095faf8d73c660401297bfa932ca46c43bd64f3 just use `-cpu` flag.

They are different neural net architectures.

train_images_path is where your images are train_hdf5 something you get as output from previous step

Merged in separate PR

Cherrypicked via https://github.com/DmitryUlyanov/Multicore-TSNE/commit/fd760979bebbb4e15389b4060c2b77ed60aa5eae as this PR has more diff than it should.

@atteneder Is there a reason not to merge this?

See this: https://github.com/atteneder/glTFast/issues/550#issuecomment-1400257136 By the way, try [Avaturn](https://avaturn.me)

Thank you David, I will give it a try today!

By the way, any plans to make a standalone hypergrad for torch? It looks like you have everything for it.