Does pytorch work with python 3.6.3, CUDA 9.0 and cuDNN 7.0 in Ubuntu 17.10?

Can I build from source using Anaconda python 3.6.3, CUDA 9.0 and cuDNN 7.0 in Ubuntu 17.10 latest stable version?

I don’t know about it’s ability on Ubuntu 17.10 but at least it works with Cuda 9.0/cuDNN 7.0/Python3.6.3.
See Pytorch on V100 GPU to check the requirements for Cuda9

I think that it depends on version of gcc compiler which has to co-work with cuda 9.0.

There are some happy customers who report successfully compiling in this configuration https://github.com/pytorch/pytorch/issues/3546#issuecomment-342895649