Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
toxik
on July 3, 2023
|
parent
|
context
|
favorite
| on:
Douglas Hofstadter changes his mind on Deep Learni...
It depends on how you define fed forward, LLMs are typically auto regressive and so can take their own previous output into consideration when generating tokens.
Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: