• 2 Posts
  • 14 Comments
Joined 1 year ago
cake
Cake day: June 17th, 2023

help-circle




  • I don’t know how tech savvy you are, but I’m assuming since your on lemmy it’s pretty good :)

    The way we’ve solved this sort of problem in the office is by using the LLM’s JSON response, and a prompt that essentially keeps a set of JSON objects alongside the actual chat response.

    In the DND example, this would be a set character sheets that get returned every response but only changed when the narrative changes them. More expensive, and needing a larger context window, but reasonably effective.










  • I suspect this is a rendering/process thing rather than an accuracy thing. Have you tested if the logic of the first one isn’t listening to the other fields?

    Thinking it through, in an ‘and’ test, you do only have to watch for the the state of the first clause, because so long as that is false, it doesn’t matter what the rest say. In the ‘or’ clause, you need to watch for the value of all of them.