Remix clone Hacker News
new
|
show
|
ask
|
jobs
Github
▲
SubQ 1.1 Card: Linear-scaling sparse attention with 98% retrieval at 12M tokens [pdf]
(
subq.ai
)
2 points
by
mitchwainer
5 hours ago
|
1 comments
▲
mitchwainer
5 hours ago
|
parent
[-]
[flagged]