Optimize Transpose->Pad when 'pads' input is not a constant #1331
+43
−1
Conversation
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
When the second input is not a constant, let's shuffle it with Split followed by Concat. There are examples of models, where this non-constant input gets constant folded anyway by a framework. Even if that's not the case. Split+Concat of 8 (or 10) element tensor should be a good trade for Transpose pair. Signed-off-by: Mateusz Tabaka <[email protected]>
|
@TomWildenhain-Microsoft this one is ready for review. Thanks. |
TomWildenhain-Microsoft
approved these changes
Feb 10, 2021
| # when the second input is not a constant, let's shuffle it with Split followed by Concat | ||
| # there are examples of models, where this non-constant input | ||
| # gets constant folded anyway by a framework. | ||
| split = self._g.make_node("Split", inputs=[node.input[1]], attr={}, output_count=trans_rank * 2) |
There was a problem hiding this comment.
Nice. When the pad is non-const, it is often constructed by concating some tensors together. This optimization might introduce a concat -> split -> concat sequence that can be further optimized. Still, it is worth it for removing a transpose and the concat -> split -> concat will be operating on very small tensors. We might want to add a new optimizer for it in the future though.
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Outdated suggestions cannot be applied.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.
When the second input is not a constant, let's shuffle it with Split followed by Concat.
There are examples of models, where this non-constant input gets constant folded anyway by a framework.
Even if that's not the case, Split+Concat of 8 (or 10) element tensor should be a good trade for Transpose pair.
Signed-off-by: Mateusz Tabaka [email protected]