Biggest takeaway: extraction of prompts seems to be complete bullshit. | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		brcmthrowaway on Dec 24, 2023 \| parent \| context \| favorite \| on: "Attention", "Transformers", in Neural Network "La... Biggest takeaway: extraction of prompts seems to be complete bullshit.

astrange on Dec 25, 2023 [–]

System prompts are (usually) just text prepended to your own prompt, and an LLM is certainly capable of reliably quoting text fed into it.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact