Add support for gradient clipping #1331
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
hturki
force-pushed
the
ht/gradient-clip
branch
from
bc3eed0
to
5cd7554
Compare
hturki
force-pushed
the
ht/gradient-clip
branch
from
5cd7554
to
7f72e23
Compare
tancik
approved these changes
Feb 1, 2023
|
i wasn't planning on it, have been testing on fairly specific usecases so far with a hash-based (but not nerfacto) architecture, and I'm not sure whether this is needed for nerfacto / the specific clipping values I'm using will generalize. For nerfacto I'd maybe start with adding weight decay per #873 and then we could figure out reasonable clipping values if training is still unstable. |
|
Gotcha 👍 |
chris838
pushed a commit
to chris838/nerfstudio
that referenced
this pull request
Apr 22, 2023
implement optional gradient clipping
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
I've found that this makes training more stable in certain cases, especially when using high learning rates