DeepFaceLab Change `tensorflow-gpu` to `tensorflow`

Hey there,

Wanted to check with you first, currently in requirements-cuda.txt we find tensorflow-gpu=2.4.0 which if I am not mistaken is legacy, and tensorflow=2.4.0 would just work as well, this is since tensorflow > 2.

I've wasted some time trying to set this up on linux, and this was one of the things that I had to change to make it work (among others), actually I used tensorflow without an specific version (installed 2.5.0) and so far so good, so it might be able to be bumped.

Let me know if you wnat a PR.

Cheers.

Jul 23 '21 16:07 Minkiu

Hi, Im working on getting this working on linux as well. Thanks for the tip, can you further detail what changes you have had to make? Hopefully will save me retreading the same ground.

Jul 24 '21 05:07 Meatfucker

Hey, so you can check the PR i did agaisnt the "linux" version of the project: https://github.com/nagadit/DeepFaceLab_Linux/pull/42

But boils down to settingup the codna environment correctly, which the current guide is not obvious about some parts and I tripped on those, let me know if that is enough.

Jul 28 '21 16:07 Minkiu

Did you ever find the answer? If so, would you mind sharing it and closing this issue?

Jun 08 '23 23:06 joolstorrentecalo

Hi @JDTCC i did create a PR with the findings, but it hasn't been merged: https://github.com/nagadit/DeepFaceLab_Linux/pull/42

I guess the issue can be closed, yes.

Jun 14 '23 09:06 Minkiu