Fix and simplify pytorch backend: array, einsum, from_numpy #1550
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
Codecov Report
@@ Coverage Diff @@
## master #1550 +/- ##
==========================================
- Coverage 90.73% 89.96% -0.76%
==========================================
Files 102 102
Lines 10414 10288 -126
==========================================
- Hits 9448 9255 -193
- Misses 966 1033 +67
Flags with carried forward coverage won't be shown. Click here to find out more.
Continue to review full report at Codecov.
|
ninamiolane
approved these changes
May 31, 2022
There was a problem hiding this comment.
Thank you for this great simplification! I can't remember exactly why einsum was this complicated, it could be to allow einsums:
- of the type
("ijk, jkl->ijl", a, b)i.e. with tensors of order 3 and more - of the type
("ijk, jkl, mjkl,->mijl", a, b, c)i.e. with 3 tensors or more - with ellipsis, i.e. of the type
("...ijk, ...jkl..., mjkl,->...mijl...", a, b, c) - a combination of the above,
and it could be that previous versions of pytorch were not allowing it but new ones do?
In any case, if tests do not fail we can keep it like this... and remember where to look for if we encounter problems.
Thanks!
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.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 PR acts specially in
arrayfortorchbackend. It avoids unnecessary conversions tonumpy.Additionally,
dtypehas been added tofrom_numpytocastwhen required. It needs to be extended to the other backends for compatibility, but it will play a role whenscipyis used withtorchortensorflow(to ensure precision is kept).einsumis also simplified. @ninamiolane the tests are not failing with this simplification, but I'm not sure if I'm missing something (I was not able to understand whyeinsumhad such a complex wrapper). If you have an example that justifies the wrapper, please let me know so I revert the changes there.It closes #1537 and starts paving the path to better performance in GPU.