Setting up Linux computer for ML (cuda drivers)

As I've setup several desktop computers with ubuntu for on-prem ML, sometimes installing certain nvidia drivers on ubuntu can cause the graphics driver to throw an error while booting and then will give you a black screen. Here's a setup of commands to fix it:

1. Reboot machine and go to recovery mode using grub (Advanced Ubuntu)

    a. Enable Networking

    b. Drop to Root

2. sudo nvidia-uninstall #To remove all  remaining nvidia drivers, to ensure clean install

3. sudo add-apt-repository ppa:graphics-drivers/ppa

4. sudo apt-get update

5. sudo apt install nvidia-418 #Could try a newer one, but this was the most recent driver that worked for ML for me on my 2080Ti card

6. Reboot - this should've fixed it.

7. nvidia-smi #will show you if cuda and gpu driver version