← Back to context

Comment by saltsucker

16 hours ago

Comments later in thread say markdown works just as fine and that it’s more important to organize your plan into sections.

Also just think about it, why would a model trained on the world’s corpus of text (that isnt formatted in xml) perform better with XML? It would be a better study if that post tested markdown, org, xml, json, etc. 10 times to see if their is a difference

A year or so ago XML worked more reliably for long-lived prompt instructions. Now it is cargo culting.

XML consistently performed better than markdown and JSON in all evals I've ever seen on any model, except for a couple very specific ones.