Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
loading thread…
3 ms
·
Native Sparse Attention: Hardware-Aligned and Natively Trainable
2 points
by
teepo
2y ago