Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
energy123
on March 7, 2025
|
parent
|
context
|
favorite
| on:
Some thoughts on autoregressive models
Yann LeCun's prediction was empirically refuted. He says that the longer LLMs run, the less accurate they get. OpenAI showed the opposite is true.
mrfox321
on March 7, 2025
[–]
They didn't show this, they just increased the length where accuracy breaks down.
energy123
on March 7, 2025
|
parent
[–]
Explain? OpenAI showed the new scaling law in December 2024 that performance keeps increasing proportional to ln(N reasoning tokens)
mentalgear
on March 7, 2025
|
root
|
parent
[–]
link?
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
