upgrade tensorflow version and adapt server.py

[francoislamiedepain]

Hello,

I have upgraded tensorflow version and related python libraries listed in requirement.txt.

I have also upgraded python image version in Dockerfile.

And eventually i have added a bit more infos in README as i have face some issues with new node version.

Test on local and with docker-compose and everything is working fine ;)

[johndpope]

nice @francoislamiedepain - did you get stylegan2 running with tf2? NVlabs seem to be jumping ship to pytorch. but.... I wonder if it's not too hard to do. I had a go at auto migrating (https://www.tensorflow.org/guide/upgrade) this https://github.com/previtus/GAN_explorer and seems like it just needs a few more tweaks to get working. (though I got stuck)

NVlabs/stylegan2-ada#52

UPDATE
might be better to use this based image for tensorflow2
docker pull nvcr.io/nvidia/tensorflow:20.12-tf2-py3

check here for latest
https://docs.nvidia.com/deeplearning/frameworks/tensorflow-release-notes/running.html

[francoislamiedepain]

Hi @johndpope , i did not try stylegan2 running with tf2 but it's still seems to be not a easy task... :

"We recommend TensorFlow 1.14, which we used for all experiments in the paper, but TensorFlow 1.15 is also supported on Linux. TensorFlow 2.x is not supported. " (https://github.com/NVlabs/stylegan2#requirements).

Yes definitely a better approach to use tf2 based image.

Still need more work to get everything clean.

Thanks for the links !