Mistral model too chatty

[kltye]

I'm using Ollama with the Mistral model, and the responses can sometimes be too verbose when I ask for sensor statuses. It particularly likes to tell me how it derived the information ("The front door is locked, as indicated by sensor..." or "The washer is currently idle sensor.washer...") I've added prompts like telling it to respond concisely, not to tell me where the data comes from etc etc. It has somewhat mitigated the issue, but it still likes to be too verbose pretty often.

On another note, is it possible to use the Home model with Ollama and have it be GPU accelerated?