autumn panther
Hello there, just finished setting up a complete assist setup with LLM integration.
It does work for turning on/off lights as shown in the example videos but this concrete example did not work at all.
I saw this JSON response from the LLM a couple times already, but it seems really broken.
Am I doing something wrong or is this currently still expected?
autumn panther
A bit hard to find up to date information on this, is there a way to debug the prompts and HASS api calls it does?
autumn panther
Is this answer from LLM? Looks like HASSIL to me.
so to answer this, I have no clue.
the CPU usage jumped and I got ollama logs so I'm assuming it comes from the LLM
hollow orbit
This is usually a sign that either the model isn't doing function calling right (is it the instruct version of the model?) or you are surpassing the context limit for your ollama instance. Might want to check the logs on the ollama side to see if the request or context is being truncated.
Also not sure if llama3.2 performs well for this
autumn panther
Also not sure if llama3.2 performs well for this
probably not, was hoping to get by without GPU support with my i3-14100
but even that model takes quite a long time to compute
