[Ready] Batched cholesky decomposition #14017
Conversation
vishwakftw
force-pushed
the
batch_cholesky
branch
from
9fd25cc to
49e4efd
Compare
test/test_cuda.py
Outdated
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
vishwakftw
changed the title
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@ezyang has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
|
@ezyang I have addressed comments and resolved merge conflicts too. |
|
@pytorchbot retest this please |
vishwakftw
force-pushed
the
batch_cholesky
branch
from
1313d1b to
3987642
Compare
|
Build failures are unrelated. |
facebook-github-bot
left a comment
facebook-github-bot
left a comment
There was a problem hiding this comment.
@ezyang is landing this pull request. If you are a Facebook employee, you can view this diff on Phabricator.
Summary: Implements batching for the Cholesky decomposition. Performance could be improved with a dedicated batched `tril` and `triu` op, which is also impeding autograd operations. Changes made: - batching code - tests in `test_torch.py`, `test_cuda.py` and `test_autograd.py`. - doc string modification - autograd modification - removal of `_batch_potrf` in `MultivariateNormal`. Pull Request resolved: pytorch/pytorch#14017 Differential Revision: D13087945 Pulled By: ezyang fbshipit-source-id: 2386db887140295475ffc247742d5e9562a42f6e
| A = cast(random_symmetric_pd_matrix(n, *batch_dims)) | ||
| cholesky_exp = torch.stack([m.cholesky(upper=upper) for m in A.reshape(-1, n, n)]) | ||
| cholesky_exp = cholesky_exp.reshape_as(A) | ||
| print(torch.cholesky(A, upper=upper)) |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
4 participants
Implements batching for the Cholesky decomposition.
Performance could be improved with a dedicated batched
trilandtriuop, which is also impeding autograd operations.Changes made:
test_torch.py,test_cuda.pyandtest_autograd.py._batch_potrfinMultivariateNormal.