Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
whimsicalism
on Jan 25, 2025
|
parent
|
context
|
favorite
| on:
DeepSeek-R1: Incentivizing Reasoning Capability in...
because they’re able to pass signal on tons of newly generated tokens based on whether they result in a correct answer, rather than just fitting on existing tokens.
it’s on the path to self play
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
it’s on the path to self play