Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
pityJuke
on Dec 1, 2023
|
parent
|
context
|
favorite
| on:
Are Open-Source Large Language Models Catching Up?
Is it hallucinating (whether that be through sheer chance, or trained to think it is GPT), or is it pointing at the wrong place?
https://imgur.com/a/YOF6szw
or
https://imgur.com/a/fkgkfRO
zwily
on Dec 1, 2023
[–]
Probably just trained on lots of GPT-4 output.
moffkalast
on Dec 1, 2023
|
parent
|
next
[–]
Apparently trained on lots of refusals too, speaks to the high competence of whoever was setting up the dataset. It's one string regex to filter them out and get more performance for fucks sake.
pityJuke
on Dec 1, 2023
|
parent
|
prev
[–]
Oh, right, I remember hearing that was a technique to train LLMs. Interesting that it impacts it in such a way.
Join us for
AI Startup School
this June 16-17 in San Francisco!
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
