At first glance this looks very similar to just adding the contrastive prompts t... | Hacker News

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

		rgbrgb on Feb 18, 2024 \| parent \| context \| favorite \| on: Representation Engineering: Mistral-7B on Acid At first glance this looks very similar to just adding the contrastive prompts to the beginning of the system prompt to "prepare" the logits. What am I missing?

gsuuon on Feb 19, 2024 [–]

I guess since the control vectors are applied at every layer, it becomes impossible to override (e.g. if you're consuming user-generated prompts)

Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10
Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact