Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
curious_cat_163
on June 24, 2024
|
parent
|
context
|
favorite
| on:
Detecting hallucinations in large language models ...
It’s a pretty clever idea: “check” if the model answers “differently” when asked the same question again and again and again.
“checking” is being done with another model.
“differently” is being measured with entropy.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
“checking” is being done with another model.
“differently” is being measured with entropy.