Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
quickthrower2
on July 30, 2023
|
parent
|
context
|
favorite
| on:
Computer scientists discover limits of gradient de...
Gradient descent also allows batching, which reduces the memory requirements, can the other methods support that?
nerdponx
on July 31, 2023
[–]
As far as I know, this is the #1 gradient descent superpower that makes it the preferred choice above all others. I don't think e.g. L-BFGS supports batching, I've certainly never seen it.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
