Open Deepak- opened 8 years ago
Use Virtualbox, and run ubuntu. Thats the easiest way.
Ah yes, thanks @heil-stalin. So it's definitely not possible to do this on a Windows machine without using Virtualbox?
The simplest way for you probably would be using docker image with torch, but it still uses VirtualBox for visualization.
Thank you for the tip @susloparovdenis I will look into that
@Deepak- neural-style depends on Torch7; installing Torch7 is not officially supported and probably won't be supported in the future; see https://github.com/torch/torch7/issues/150.
Some brave people have gotten torch to build and run on windows, but it's probably not worth the effort; as others have suggested I'd recommend virtualizing with something like VirtualBox.
@jcjohnson Thank you! I think I know enough now :) Virtualizing seems the way to go yes.
@jcjohnson Thank you! I think I know enough now :) Virtualizing seems the way to go yes.
Only downside to a VB in windows is no graphics card support. Very little resources are lost when using CPU, though.
@heil-stalin Indeed. Takes about 3 hours now with all the default settings ;)
Well, I have dual hexa-core xeons from 2010. Neural style only uses one core. So, mine takes 6+ hours. It'd be nice if it could use all 12 cores
@heil-stalin Wow that's crazy. Having a nice GPU would definitely be welcomed from my end as well!
Follow these instructions (use CUDA 7.0, not current 7.5). https://github.com/Syntopia/deepdream-windows-installation
@n1ckfg thanks, will check it out!
@n1ckfg What about Torch? I used this method to get DeepDream running, but I'm not sure what I'd do about torch.
@Taconut This is getting kind of off-topic, but there's a Caffe version of the algorithm that works on Windows: https://github.com/fzliu/style-transfer
Dear all,
Is it possible to run neural-style on Windows?
Thanks in advance.