News for hackers. Stuff that matters.

A comment on How Distillation Makes AI Models Smaller and Cheaper. Parent comment

Re:How Distillation Makes AI Models Smaller and Cheaper by anonymoushn on Thursday July 24, 2025 @08:03AM (#44669654)
"subliminal learning" does not even work for use cases like distilling o1 to R1 because they do not share a base model
[ Parent | on HN ]
Re:How Distillation Makes AI Models Smaller and Cheaper by pyman on Thursday July 24, 2025 @09:13AM (#44670289)
Who's talking about that?

[Edit] My bad, I thought I was commenting on Anthropic's article

[ Parent | on HN ]
Re:How Distillation Makes AI Models Smaller and Cheaper by anonymoushn on Thursday July 24, 2025 @09:17AM (#44670325)
i replied to a comment by the hacker news user called pyman which claimed incorrectly that distillation was repackaged as "subliminal learning". so if you are asking me, who is talking about subliminal learning, which is unrelated to the topic of the article, the answer is that the hacker news user called pyman is doing that.
[ Parent | on HN ]
Re:How Distillation Makes AI Models Smaller and Cheaper by pyman on Thursday July 24, 2025 @09:42AM (#44670595)
Ah you are right, I was commenting on this article:

https://alignment.anthropic.com/2025/subliminal-learning/

[ Parent | on HN ]