'name 'Unet34' is not defined' - How to run code on CPU?

[AlexanderKozhevin]

As far as I got it right, this class is related to CUDA.
But how can I run it on CPU ?
I disabled this line of code.

#torch.cuda.set_device(0)

[jantic]

Soo...I tried this before as well, namely because i wanted to see if I could run bigger images (more memory with CPU of course).

The way to really force CPU is putting this up top in the notebook before the pytorch related imports:

import os
os.environ['CUDA_VISIBLE_DEVICES'] = ''

When I tried that myself, it seemed to have done the trick, but then the processing was super slow and I just abandoned the effort in the meantime. The reason why I abandoned it was because I looked up the CPU performance issue and the man himself (Soumith Chintala) said this:

@Zhangchenkai you will see great improvement if you build from source from these instructions: https://github.com/pytorch/pytorch#from-source

The instructions will build and link with mkl-dnn and that will show you remarkable improvements to what you see.

pytorch/pytorch#9873

I think that's probably the issue I saw. I didn't go ahead and build from source because it wasn't pressing and I'm a bit lazy. You could try it...I just don't know for certain if that's actually the issue.

Please, if you're so inclined and do try it out, please let me know! We can put it in the readme then because that'll probably be helpful to a lot of people.