Add simple argument-only lifted nn.grad function. #3463
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
cgarciae
reviewed
Nov 6, 2023
cgarciae
reviewed
Nov 6, 2023
cgarciae
reviewed
Nov 6, 2023
levskaya
force-pushed
the
vjp_fix
branch
2 times, most recently
from
ea21918
to
30e54dd
Compare
cgarciae
approved these changes
Nov 6, 2023
This function only peforms a lifted value-and-grad operation with respect to the arguments of a function, and does not try to calculate gradients with respect to any variables. This mirrors the behavior of the haiku grad function, and also easily works in the multi-scope setting (external modules passed in) while avoiding the complexities associated with that case for a more general vjp.
Codecov Report
@@ Coverage Diff @@
## main #3463 +/- ##
==========================================
+ Coverage 83.56% 83.62% +0.05%
==========================================
Files 56 56
Lines 6768 6790 +22
==========================================
+ Hits 5656 5678 +22
Misses 1112 1112
|
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Outdated suggestions cannot be applied.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.
This function only peforms a lifted value-and-grad operation with respect to the arguments of a function, and does not try to calculate gradients with respect to any variables. This mirrors the behavior of the haiku grad function, and also easily works in the multi-scope setting (external modules passed in) while avoiding the complexities associated with that case for a more general vjp.