Skip to content
DeepFaceLab CentOS Linux
Branch: master
Clone or download

Latest commit

elemantalcode Merge pull request #4 from elemantalcode/ISSUE-1
Removed patch as main code includes experimental codec
Latest commit 70f2e00 Mar 11, 2020


Type Name Latest commit message Commit time
Failed to load latest commit information.
scripts Added no-preview script for CLI-based environments Feb 8, 2020
LICENSE Create LICENSE Feb 10, 2020 Removed patch as main code includes experimental codec Mar 10, 2020
setup.bash Removed patch as main code includes experimental codec Mar 10, 2020

DeepFaceLab CentOS Linux

Note: This installation method works for CentOS 7.x - your milage may vary.

This is a simple addition to DeepFaceLab to allow the latest code to be run directly on Linux.

Install requirements

Download the NVIDIA cuDNN RPM. note: You need to register and make sure you download the specific version libcudnn7-

Manual method

Install the rest of the requirements as your favourite user (root is used by default and as example):

yum -y install epel-release
wget -O /etc/yum.repos.d/cuda.repo
yum -y update
yum -y install python3-3.6.8 git cuda-toolkit-10-0.x86_64 ffmpeg ffmpeg-devel libSM libXext libXrender-devel libXrender gcc gcc-c++ 
yum -y install kernel-devel make cuda-libraries-10-0.x86_64 nvidia-driver-cuda.x86_64 nvidia-driver-cuda-libs.x86_64 

cd ~ && git clone
cd ~ && git clone

cd ~/DeepFaceLab && cp -advpR ../dfl/scripts .
cd ~/DeepFaceLab && python3.6 -m pip install -r requirements-cuda.txt
yum -y install /root/libcudnn7-

Scripted method

Run the setup.bash script

sudo setup.bash

Post-install check

Once installed, you should be able to run DeepFaceLab as per usual:

cd ~/DeepFaceLab/scripts

Quick FAQ

Q: Why CentOS?

A: It's stable, close to RHEL and doesn't suffer the same fate as CoreOS ... yet.

Q: Why not docker?

A: Why abstract yet again when you don't have to?

Q: Why not use Anaconda3 like DeepFaceLab for Linux

A: Different approaches to solve the same challange. I personally don't like adding frameworks/structures that I don't intend on making good use of. For python developers, Anaconda makes sense - for me, I just want to run DeepFaceLab under Linux with minimal hassle.

Q: Anything else particular noteworthy?

A: You need some Linux knowledge.

You can’t perform that action at this time.