Skip to content

Comment on “Disregard That” Attacksparent

Comments

This doesn't work for the tasks people are worried about because they want to lean on the generalization of the model + tool calling.

What you're describing is also already mostly achieved by using constrained decoding: if the injection would work under constrained decoding, it'll usually still work even if you SFT heavily on a single task + output format

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.