Y
Hacker News
new
|
ask
|
show
|
jobs
DeepSeek's Multi-Head Latent Attention
(
liorsinai.github.io
)
4 points
by
the_origami_fox
467 days ago
1 comments
fspeech
467 days ago
Matrix absorption is unnecessary. What is needed is the order of multiplication associates towards the direction of the absorption. This and the modified Rope are needed to make the caching work.
link