-
Notifications
You must be signed in to change notification settings - Fork 10.6k
[AutoDiff] Reabstraction thunk differentiation. #28570
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
Merged
Changes from all commits
Commits
Show all changes
8 commits
Select commit
Hold shift + click to select a range
85c828e
[AutoDiff] Differentiate partially-applied reabstraction thunks.
dan-zheng 35a6fe3
Refactor tests to avoid orthogonal issue TF-1033.
dan-zheng 8f71652
Merge branch 'tensorflow' of github.com:apple/swift into TF-201
dan-zheng c7c0ebe
Minor renaming for clarity.
dan-zheng 8f0659d
Address review feedback.
dan-zheng f6c56f1
Reference TF-1036 optimizations.
dan-zheng 9d13af1
Merge branch 'tensorflow' of github.com:apple/swift into TF-201
dan-zheng e870b6d
Fix test based on upstream changes.
dan-zheng File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
This file contains hidden or 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
Oops, something went wrong.
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.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It would be helpful to have a documentation comment explaining what the derivative function of a reabstraction thunk is.
If I understand correctly, it is "a function that, when partially applied to the original function, becomes the derivative function of the reabstracted function." Or maybe instead it "becomes a reabstraction of the derivative function". I haven't thought carefully enough to figure out which one it is, or if there is a difference, but maybe there are situations where the difference is important.
Another question I have is: Why does the derivative function of the reabstraction function take the original function as argument rather than than the derivative function? That might be a possible alternative with advantages or disadvantages. I haven't thought enough about it to determine whether it's actually possible or whether it's a better alternative, I'm just wondering if you have. A comment somewhere about why we decided one way or the other would be useful.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Added doc comment for
SILFunctionType::getAutoDiffDerivativeFunctionTypein 6eb7a07:Also added comment in
reapplyFunctionConversionin Differentiation.cpp explainingpartial_applyreapplication for reabstraction thunk derivatives.I believe your understanding is correct. I made a short write-up about the reabstraction thunk differentiation approach.
That's a good point! A reabstraction thunk derivative only needs the original function's corresponding derivative, not any other
@differentiablefunction components. I think that approach is more efficient and semantically clear, so I'll pursue it now.There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually, to focus on retrodiff as a priority, how about deferring optimizations and using the current approach for now? TF-1036 tracks reabstraction thunk derivative generation optimizations.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
sounds good
nice doc comment!