Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
patrakov
on Aug 8, 2023
|
parent
|
context
|
favorite
| on:
GPT-4 can't reason
This non-deterministic sampling is not only for users to explore the space of responses. Without this, the LLM itself is prone to generate too-repetitive text.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
