Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
|
redlock's submissions
login
1.
New deepseek paper: Natively Trainable Sparse Attention mechanism
(
twitter.com/deepseek_ai
)
5 points
by
redlock
6 months ago
|
past
|
1 comment
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
