I don’t have an LLM tab on my device; I only have this options on the main AI settings page
How was your instance installed? Are you running latest of Discourse? Can you try in safe mode without any customizations?
I tried safe mode, but nothing worked. The software was installed on a Raspberry Pi 5 running Raspberry Pi OS Lite with Docker. I am currently using version 3.6.0.beta3-latest.
You are going to need to upgrade to latest as a minimum. Discourse v2025.11.0-latest is what you should be on.
1 like
thanks I will do that now
ok how do I go about upgrading to the latest
If it is a standard docker based install ./launcher rebuild app will do the trick.
2 likes
I am all upto date now, thanks friend, but how do I go about now using one of my self hosted models on my rasberry pi with Ollama if that even possable
2 likes
While possible, hosting both an entire LLM as well as Discourse on the Pi will most likely leave you with an unsatisfactory performance for both.
2 likes
Falco actually does this
1 like
I tried running Ollama on a Pi. It took so long it gave up trying to connect. Unless you have a good deal of resources, running Ollama on a 4GB RAM Pi isn’t the best idea. I’m not sure if swap would help much.
On a similar but off-topic note, in my experience, running a Discourse dev env on a 4GB RAM Pi with no swap on a SD Card doesn’t work. But that’s neither here nor there.
2 likes