How does chat exactly work when you mention two or more part of the code?
I want to say look at X then do Y
but when you type all the code goes to the top.
and the text and code does respectively in order i want to type.
doesn’t it mess the whole chat and talking with LLM?
1 Like
they can make it to index them, and can be mentioned like:
- @1 or @snippet1
- @2 or @snippet2
- @N or @snippetN
2 Likes
yeah when you mention Documents it’s alright and it’s like this:
but when you mention the code all of them goes at the top and it’s messy:
1 Like
That’s because the mention is indexed (or embedded) like the codebase.
but when the file/snippet gets at the top, that means the content has been copied into your prompt (not indexed).
yeah got that
but embedding will not work in this situation
too much code too much request and also take some times.
if they just structure the chat the way we type it i guess that would fix the problem.
writing
@ mentioned code
continue writing
@ mentioned code
1 Like
Yes, the main problem here is that they stack to the top. and that’s less controllable.
Agree.
1 Like
Exactly. and i think if they get sent in that way into a LLM, then the quality of the answer wouldn’t be that good COMPARE to when you structure the request in the best way, right!?
1 Like
Yeah this seems like a reasonable request - unsure if we will have the bandwidth for doing this in the next week, but will push for it
2 Likes