Skip to content

Automated deployment of Deeplabcut#5

Open
j-i-l wants to merge 13 commits into
mainfrom
4-role-for-setting-up-cuda-support
Open

Automated deployment of Deeplabcut#5
j-i-l wants to merge 13 commits into
mainfrom
4-role-for-setting-up-cuda-support

Conversation

@j-i-l

@j-i-l j-i-l commented Dec 18, 2024

Copy link
Copy Markdown
Member

This implements an ansible role to setup and configure a vanilla ubuntu machine (with nvidia gpu) for DeepLabCut

@j-i-l j-i-l self-assigned this Dec 18, 2024
@j-i-l j-i-l linked an issue Dec 18, 2024 that may be closed by this pull request
@j-i-l

j-i-l commented Dec 18, 2024

Copy link
Copy Markdown
Member Author

We might want to rename the role as it installs DeepLabCut3.

Or we split it up into several roles:

  • setting up conda
  • installing pytorch
  • installing deeplabcut

Up to installing conda could be a reasonable split. Afterwards pytorch could also be installed with RocM on AMD hardware. Also deeplabcut works with pytorch only from 3.0 on (see DeepLabCut/DeepLabCut#2613) and needs tensorflow for earlier versions...

@j-i-l

j-i-l commented Dec 27, 2024

Copy link
Copy Markdown
Member Author

This is quite the mess for a declarative setup.

the constraint tf <=2.10 applies also for ubuntu (not 2.12 as indicated. tf 2.10 limits us to python <=3.10.

We should adapt our default parameter values accordingly!

@j-i-l

j-i-l commented Jan 1, 2025

Copy link
Copy Markdown
Member Author

A better approach might be to directly checkout a repo that contains some analysis performed with DeepLabCut and fetch the dlc version form the repo's requirements.txt.

@j-i-l j-i-l force-pushed the 4-role-for-setting-up-cuda-support branch from cb613ad to e91bbac Compare September 18, 2025 10:46
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

role for setting up cuda support

1 participant