Y
Hacker News
new
|
ask
|
show
|
jobs
reilly3000
2 days ago
[
-
]
It has to be 2^n nodes and limited to one per attention head that the model has.
Please enable JavaScript to continue using this application.