alt
Hacker News
bigyabai
•
yesterday at 9:33 PM
•
0 replies
•
view on HN
>90% of inference hardware is faster if you run an MOE model.
alt Hacker News