Dataflow matrix machines (by Anhinga anhinga) (
dmm
) wrote
2021-08-17 03:48 pm (UTC)
no subject
Talk 4 - very interesting
Replacing rules for coarse-grained evolution so that it is precise(that is, equal to coarse-graining of the high-res simulation)
Trained from high-res simulations of small pieces
"Put in as little machine learning as possible and see how far we can get"
He likes doing things in JAX:
https://github.com/google/jax-cfd
Multi-step loss is essential
"Some of the problems mentioned at the beginning are still intractable, but every 1000x acceleration helps ;-) " - I love this way to formulate it
(
17 comments
)
Post a comment in response:
From:
Anonymous
This account has disabled anonymous posting.
OpenID
Identity URL:
Log in?
Dreamwidth account
Account name
Password
Log in?
If you don't have an account you can
create one now
.
Subject
HTML doesn't work in the subject.
Formatting type
Casual HTML
Markdown
Raw HTML
Rich Text Editor
Message
[
Home
|
Post Entry
|
Log in
|
Search
|
Browse Options
|
Site Map
]
no subject
Replacing rules for coarse-grained evolution so that it is precise(that is, equal to coarse-graining of the high-res simulation)
Trained from high-res simulations of small pieces
"Put in as little machine learning as possible and see how far we can get"
He likes doing things in JAX: https://github.com/google/jax-cfd
Multi-step loss is essential
"Some of the problems mentioned at the beginning are still intractable, but every 1000x acceleration helps ;-) " - I love this way to formulate it