Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
kolinko
on May 1, 2024
|
parent
|
context
|
favorite
| on:
Kolmogorov-Arnold Networks
I may be wrong but with midern llms biases aren’t really used any more.
tripplyons
on May 1, 2024
[–]
From what I remember, larger LLMs like PaLM don't use biases for training stability, but smaller ones tend to still use them.
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:
