Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
unsatchmo
5 months ago
|
parent
|
context
|
favorite
| on:
Apple M3 Ultra
You're correct about the weights: each machine could in fact store all of the weights. However I think you still have to transfer the activations and the KV-Cache while performing inference.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: