Sampling repeatedly gives them an estimate of the probability distribution in an... | Hacker News

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

		hooloovoo_zoo 78 days ago \| parent \| context \| favorite \| on: Alleged Distillation Attacks by DeepSeek, Moonshot... Sampling repeatedly gives them an estimate of the probability distribution in any case though.

hooloovoo_zoo 78 days ago [–]

That would be an interesting paper actually; what is the optimal sampling technique given you only have access to the token outputs. Surely someone has already done it.

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