-
Notifications
You must be signed in to change notification settings - Fork 26.3k
Updated derivatives for complex mm, mv, ger, bmm, triangular_solve #45737
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
Conversation
Enable gradgradcheck for complex cholesky
💊 CI failures summary and remediationsAs of commit f0908d6 (more details on the Dr. CI page): 💚 💚 Looks good so far! There are no failures yet. 💚 💚 This comment was automatically generated by Dr. CI (expand for details).Follow this link to opt-out of these comments for your Pull Requests.Please report bugs/suggestions on the GitHub issue tracker or post in the (internal) Dr. CI Users group. This comment has been revised 16 times. |
|
@IvanYashchuk this PR looks good overall. Sorry I didn't get a chance to finish working on #42553 because of complex autograd issues. I would try to work on it this week. In case, it's blocking a lot of other work, would you like to take over #42553? |
Correctness of the derivatiev rule for bmm is being implicitly tested in triangular_solve for now.
We should remember to re-enable them once batched matmul for complex is supported on cuda
facebook-github-bot
left a comment
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.
@anjali411 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
Codecov Report
@@ Coverage Diff @@
## master #45737 +/- ##
=======================================
Coverage 68.20% 68.20%
=======================================
Files 410 410
Lines 53453 53453
=======================================
Hits 36458 36458
Misses 16995 16995 Continue to review full report at Codecov.
|
| 'cosh', '__rmul__', 'sgn', 'abs', 'dot', 'vdot', 'tensor_split', | ||
| 'matmul', 'bmm', 'mv', 'ger', 'diagonal', ] + separate_complex_tests | ||
|
|
||
| # this list corresponds to cases that are not currently implemented |
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.
nit - not currently implemented for complex tensors
anjali411
left a comment
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.
LGTM thanks @IvanYashchuk
facebook-github-bot
left a comment
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.
@anjali411 has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
|
@anjali411 merged this pull request in 528158a. |
This PR updates derivatives for a few functions so that
gradgradcheckfortorch.choleskyis passed (ref).Some tests (that call to
bmm_cuda) fail with withRuntimeError: _th_bmm_out not supported on CUDAType for ComplexDoubleuntil PR #42553 is merged.
Ref. #33152