N
Hacker Next
new
past
show
ask
show
jobs
submit
login
▲
Frontier Pre-Training at Microsoft AI
(
microsoft.ai
)
30 points by
psoulos
7 hours ago
|
3 comments
add comment
Rendered at 23:14:19 GMT+0000 (Coordinated Universal Time) with Cloudflare Workers.
adroitboss 3 hours ago
[-]
Wait, I don't get it. Is the model not open? If not, why are they comparing it to Open Weight models instead of the state of the art?
leothetechguy 3 hours ago
[-]
Because SOTA Closed Base Models are not available to compare to.
Bolwin 4 hours ago
[-]
Very similar to Kimi K3 it seems, but fewer active params (80b vs 104b). I don't think they mentioned what form of attention they use