LLM usage

[lovehumans]

Description

The use of the LLM is unclear.
I see that it's used in the AI chat, for discussion of the results, that one is obvious. I also assume it's being used for translation of markers into English and for matching them to the internal marker names.
However, I manually add a marker and the LLM starts "thinking" automatically without me asking it to. This seems inefficient as I'd rather hit a button and let it do that once I'm done with data entry.

System Info

• Browser: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/146.0.0.0 Safari/537.36
• Screen: 2194x1234
• Theme: dark
• AI Provider: Local AI

[elkimek]

Do you mean when adding a value to an existing marker or creating a new custom marker entirely?

[lovehumans]

Adding value to an existing marker.

[elkimek]

That's intended. When you change the data, the context changes and needs to be reinterpreted.

[lovehumans]

Doesn't this waste computation, since it's re-interpreting an incomplete dataset over and over while I'm entering data?
It seems more efficient to re-interpret once, after the user is done adding and asks for a re-interpretation.

[elkimek]

If you use the app for a lot of manual work then yes but it was never designed for that in mind.

Doable but needs well thought UX so users who do this sporadically won't be confused as the AI interpretation is interwired in many places.