Charlie Ma <[email protected]> writes:

> the state of LLM has changed a lot over the past year since that evaluation 
> write-up.

I’ve done code-reviews of current LLMs this year and at least there the
main change is that the problems have become even harder to spot, but
they are still there.

My impressions from the latest review are “just about OK, surprisingly
good, what’s it doing here?, ok, allows full takeover of the account!,
ok, …”.

If you want to try that, the main requirement I see is: never settle for
just OK. If you settle for just OK in your review, the chance that
you’ve ovorlooked something that’s egregiously wrong is very high.

> If all the documents you have are related in subject matter, you can
> create a feedback loop to help it improve.

From your experience, would you say that that applies to the Lilypond
notation reference?
https://lilypond.org/doc/v2.26/Documentation/notation/index#top

Best wishes,
Arne
-- 
Unpolitisch sein
heißt politisch sein,
ohne es zu merken.
https://www.draketo.de

Attachment: signature.asc
Description: PGP signature

Reply via email to