Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
nkrisc
8 months ago
|
parent
|
context
|
favorite
| on:
The coming knowledge-work supply-chain crisis
How would a meaningful confidence value be calculated with respect to the output of an LLM? What is “correct” LLM output?
Kinrany
8 months ago
[–]
It can be the probability of the response being accepted by the prompter
nkrisc
8 months ago
|
parent
[–]
So unique to each prompter, refined over time?
Kinrany
8 months ago
|
root
|
parent
[–]
Only unique to the promt itself, as that's the only information it has.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: