More elaboration on the "Instructions" section of Ollama documentation

[Jeepmb]

Feedback

I'd like to request more details and possibly some examples of configuring the "Instructions" section of the Ollama configuration. The current text for that section only contains the following:

Instructions for the AI on how it should respond to your requests. It is written using Home Assistant Templating.

This statement leaves a lot of questions unaswered, like:

• Does this field equate to the "System" role prompt in models like Llama3.1 (which I believe is the default)?
• What does "...written using Home Assistant Templating" mean? Is the user supposed to write the instructions here in some template format? What exactly is being specified here?
• Can we provide personal information in these instructions (like an implementation of RAG)?
• Do instructions entered here impact how the speech to text engine vocalizes its responses?
• How much text can be entered here? How large or in-depth can these instructions be?

URL

https://www.home-assistant.io/integrations/ollama/

Version

2024.9.1

Additional information

No response

[home-assistant[bot]]

Hey there @synesthesiam, mind taking a look at this feedback as it has been labeled with an integration (ollama) you are listed as a code owner for? Thanks!

Code owner commands

Code owners of `ollama` can trigger bot actions by commenting: - `@home-assistant close` Closes the feedback. - `@home-assistant rename Awesome new title` Renames the feedback. - `@home-assistant reopen` Reopen the feedback. - `@home-assistant unassign ollama` Removes the current integration label and assignees on the feedback, add the integration domain after the command. - `@home-assistant add-label needs-more-information` Add a label (needs-more-information) to the feedback. - `@home-assistant remove-label needs-more-information` Remove a label (needs-more-information) on the feedback.