Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
frogperson
36 days ago
|
parent
|
context
|
favorite
| on:
Trinity large: An open 400B sparse MoE model
What exactly does "open" mean in this case? Is it weights and data or just weights?
someotherperson
36 days ago
[–]
It's always open weights.
jetpackjoe
36 days ago
|
parent
[–]
It's never open data
jacquesm
36 days ago
|
root
|
parent
|
next
[–]
Well, it is, it's your data to begin with after all but admitting that would create some problems.
linolevan
36 days ago
|
root
|
parent
|
next
[–]
This model is sort of interesting since it seems to be using a lot of synthetic training data – but your point stands
cyanydeez
36 days ago
|
root
|
parent
|
next
[–]
So it's a rip off of a rip off, is that whats interesting?
freakynit
35 days ago
|
root
|
parent
|
next
[–]
reminds of this recent news
https://www.medianama.com/2026/01/223-nvidia-high-speed-acce...
tucnak
35 days ago
|
root
|
parent
|
prev
[–]
unless you're Ai2
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: