#Best local Home Assistant conversation agent for Raspberry Pi 5 with 4 GB RAM
1 messages · Page 1 of 1 (latest)
Use llama server but I have no idea what kinda model you can run on that pi
I run my llm on my workstation
I don't have a normal computer to run llama at all. All my computers are old and the Pi 5 is the only one that can still run it. (maybe)
Try to find a 1b model that can do function calls, but honestly at that point it'd be better to just get used to home assistant and add custom aliases to certain actions
llama3.2 normal?
Maybe
Tinyllama worked for me on a nuc but still with ~10s latency, that was the smallest model readily available I could find on llama. I ended up using openai integration in the end for better latency
Yes, but you have to pay money for it, which I don't really want to do.