from
Hacker News
Top
New
Native Sparse Attention: Hardware-Aligned and Natively Trainable
by
teepo
on 2/19/25, 1:15 PM with 0 comments