News for hackers. Stuff that matters.

A comment on Kimi-K3 Releases on HuggingFace 7/27. Parent comment

Re:Kimi-K3 Releases on HuggingFace 7/27 by naveen99 on Monday July 27, 2026 @08:14AM (#49068545)
Isn’t that distillation ?
[ Parent | on HN ]
Re:Kimi-K3 Releases on HuggingFace 7/27 by mrfox321 on Monday July 27, 2026 @08:28AM (#49068704)
No. Distillation trains on a teach model's logits or output tokens.
[ Parent | on HN ]
Re:Kimi-K3 Releases on HuggingFace 7/27 by naveen99 on Wednesday July 29, 2026 @10:53AM (#49098346)
How do you have teacher model output anything without it being output tokens (or embeddings / intermediate logits / activations)?
[ Parent | on HN ]