Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
refulgentis
24 days ago
|
parent
|
context
|
favorite
| on:
Intellect-2 Release: The First 32B Model Trained T...
Simplifying it down to "adjusting any weights is training, ipso facto this is meaningful" obscures more light than it sheds (as they noted, RL doesn't get you very far, at all)
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: