Skip to content
Better HN
Top
Best
Ask
Show
New
Jobs
Search
⌘K
SubQ 1.1 Card: Linear-scaling sparse attention with 98% retrieval at 12M tokens [pdf]
(opens in new tab)
(subq.ai)
2 points
mitchwainer
10d ago
0 comments
Save
Share
0 comments
No comments yet.
