News for hackers. Stuff that matters.

A comment on Attention Sinks in LLMs for endless fluency.

Re:Attention Sinks in LLMs for endless fluency by Tostino on Monday October 09, 2023 @02:08PM (#37823327)
So, llama.cpp already somewhat supports this: https://github.com/ggerganov/llama.cpp/issues/3440
[ Parent | on HN ]

There are no comments on this comment yet.