News for hackers. Stuff that matters.

A comment on The Illustrated DeepSeek-R1. Parent comment

Re:The Illustrated DeepSeek-R1 by researchers on Monday January 27, 2025 @06:02PM (#42846826)
Tuning for qualitative outcomes is pretty much solved via RLHF/DPO (what this post calls "preference tuning"). Right?
[ Parent | on HN ]

There are no comments on this comment yet.