Skip to content
This repository was archived by the owner on Jun 9, 2025. It is now read-only.

Conversation

@labay-adam
Copy link

Update the Dockerfile and associated setup script to reflect the changes made in #77
Principal change was the use of pyenv to force Python 3.11.10 in a world when 3.12 is the norm. Updated the base image, which also required changing the default UID/GID.

@JarodMica
Copy link
Owner

Thanks for the PR @labay-adam!

Any reason to move to 12.6.0 and 24.04 ubuntu - is that standard now for ubuntu?

Since I'm on 22.04, I just changed it to FROM nvidia/cuda:12.2.0-devel-ubuntu22.04 for my build, but I'm also not able to connect to localhost:7860 with my WSL2 rn and I don't feel like debugging what's happening. Docker built fine and I see no errors in the terminal - I just can't verify the inference/training.

If you didn't run into any issues, I'll merge this.

@labay-adam
Copy link
Author

I'll muck about a bit more to confirm.

@labay-adam
Copy link
Author

Looks like the issue with the server visibility was the lack of an GRADIO_SERVER_NAME env variable. While I was there, though, I cleaned out some misusage of pyenv and reverted the base image back to nvidia/cuda:12.2.0-devel-ubuntu22.04. Worked a treat on an Arch host, but am happy to do the diligence on some other hosts as needed.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants