Remote (LAN) machine learning guide ?

[LeFanch]

From my understanding, it is possible to have a remote machine learning server on LAN, i.e. a strong PC with GPU/CUDA and a lightweight immich server.
In my case, it's a NAS or RPi for server, and a docker/WSL2 PC for ML
(I've already validated that CUDA works with docker WSL2).

But I don't understand how to configure them make the immish server use another ML container in LAN.
I don't see ports number in the machine learning docker compose, and can't find the documentation.
I'm sorry to ask that as it seems to be writtern somewhere in doc or obvious or for a docker user.
Thank you.

[bo0tzz]

https://immich.app/docs/guides/remote-machine-learning

[LeFanch]

I missed the .yml file to be added, thank you for the very fast response ! :)