from Hacker News

Apple puts "do not hallucinate" into prompts, it works

by djhope99 on 8/7/24, 12:04 AM with 5 comments

• by ggm on 8/7/24, 1:22 AM

  I'd like a rational explanation of how the LLM interprets "don't hallucinate" -Is it perhaps "translated" internally to the functional equivalent of a higher confidence check on output?

  Otherwise, I think it's baloney. I know there is not a simple linear mapping from plain english to the ML, but the typed word clearly is capable of being parsed and processed, its the "somehow" I'd like to understand better. What would this do the interpretation of paths through weights.

  Pretty much 'citation needed'

• by TillE on 8/7/24, 1:11 AM

  Everything about prompt engineering is just the voodoo chicken.

  https://wiki.c2.com/?VoodooChickenCoding

• by unlisted7347 on 8/7/24, 8:38 AM

  Interestingly, negative prompts for stable diffusion (like "deformed hands") has similar effect. How LLM decides what's hallucinations? Mayhaps, it double checks itself? But probably it became self-aware.

• by sva_ on 8/7/24, 12:35 AM

  X doubt