Improve numerical stability of `torch.distributions.wishart.Wishart` by nonconvexopt · Pull Request #72059 · pytorch/pytorch

nonconvexopt · 2022-01-31T14:32:42Z

Maintanance of #70377
Multiple modifications of the merged initial implementation of Wishart distribution.
cc @neerajprad

Key modifications:

torch/distributions/wishart.py: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the eps value paired to each torch.dtype
test/distributions/test_distributions.py: Test Wishart distribution implementation in numerically unstable zones, i.e df values are at ndim - 1 < df < ndim where ndim is the one dimenstion of the Wishart parameter & sample matrix.

pytorch-bot · 2022-01-31T14:32:46Z

CI Flow Status

⚛️ CI Flow

Ruleset - Version: v1
Ruleset - File: https://github.com/nonconvexopt/pytorch/blob/ec4366618d9ac34e26b451e48cceebf368dbe404/.github/generated-ciflow-ruleset.json
PR ciflow labels: ciflow/default
Add ciflow labels to this PR to trigger more builds:

Workflows	Labels (bold enabled)	Status
Triggered Workflows
linux-binary-conda	`ciflow/binaries`, `ciflow/binaries_conda`, `ciflow/default`	✅ triggered
linux-binary-libtorch-cxx11-abi	`ciflow/binaries`, `ciflow/binaries_libtorch`, `ciflow/default`	✅ triggered
linux-binary-libtorch-pre-cxx11	`ciflow/binaries`, `ciflow/binaries_libtorch`, `ciflow/default`	✅ triggered
linux-binary-manywheel	`ciflow/binaries`, `ciflow/binaries_wheel`, `ciflow/default`	✅ triggered
linux-bionic-py3.7-clang9	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/noarch`, `ciflow/trunk`, `ciflow/xla`	✅ triggered
linux-docs	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/docs`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
linux-vulkan-bionic-py3.7-clang9	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`, `ciflow/vulkan`	✅ triggered
linux-xenial-cuda11.3-py3.7-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
linux-xenial-cuda11.3-py3.7-gcc7-bazel-test	`ciflow/all`, `ciflow/bazel`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
linux-xenial-py3-clang5-mobile-build	`ciflow/all`, `ciflow/default`, `ciflow/linux`, `ciflow/mobile`, `ciflow/trunk`	✅ triggered
linux-xenial-py3-clang5-mobile-custom-build-static	`ciflow/all`, `ciflow/default`, `ciflow/linux`, `ciflow/mobile`, `ciflow/trunk`	✅ triggered
linux-xenial-py3.7-clang7-asan	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/sanitizers`, `ciflow/trunk`	✅ triggered
linux-xenial-py3.7-clang7-onnx	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/onnx`, `ciflow/trunk`	✅ triggered
linux-xenial-py3.7-gcc5.4	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
linux-xenial-py3.7-gcc7	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
linux-xenial-py3.7-gcc7-no-ops	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
pytorch-linux-xenial-py3-clang5-android-ndk-r19c-gradle-custom-build-single	`ciflow/all`, `ciflow/android`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
pytorch-linux-xenial-py3-clang5-android-ndk-r19c-gradle-custom-build-single-full-jit	`ciflow/all`, `ciflow/android`, `ciflow/cpu`, `ciflow/default`, `ciflow/linux`, `ciflow/trunk`	✅ triggered
win-vs2019-cpu-py3	`ciflow/all`, `ciflow/cpu`, `ciflow/default`, `ciflow/trunk`, `ciflow/win`	✅ triggered
win-vs2019-cuda11.3-py3	`ciflow/all`, `ciflow/cuda`, `ciflow/default`, `ciflow/trunk`, `ciflow/win`	✅ triggered
windows-binary-libtorch-cxx11-abi	`ciflow/binaries`, `ciflow/binaries_libtorch`, `ciflow/default`	✅ triggered
windows-binary-libtorch-pre-cxx11	`ciflow/binaries`, `ciflow/binaries_libtorch`, `ciflow/default`	✅ triggered
windows-binary-wheel	`ciflow/binaries`, `ciflow/binaries_wheel`, `ciflow/default`	✅ triggered
Skipped Workflows
caffe2-linux-xenial-py3.7-gcc5.4	`ciflow/all`, `ciflow/cpu`, `ciflow/linux`, `ciflow/trunk`	🚫 skipped
docker-builds	`ciflow/all`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-arm64	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-arm64-coreml	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-arm64-custom-ops	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-arm64-full-jit	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-arm64-metal	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-x86-64	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-x86-64-coreml	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
ios-12-5-1-x86-64-full-jit	`ciflow/all`, `ciflow/ios`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
libtorch-linux-xenial-cuda10.2-py3.7-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/libtorch`, `ciflow/linux`, `ciflow/trunk`	🚫 skipped
libtorch-linux-xenial-cuda11.3-py3.7-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/libtorch`, `ciflow/linux`, `ciflow/trunk`	🚫 skipped
linux-bionic-cuda10.2-py3.9-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/linux`, `ciflow/slow`, `ciflow/trunk`	🚫 skipped
linux-bionic-rocm4.5-py3.7	`ciflow/linux`, `ciflow/rocm`	🚫 skipped
linux-docs-push	`ciflow/all`, `ciflow/cpu`, `ciflow/linux`, `ciflow/scheduled`	🚫 skipped
linux-xenial-cuda11.3-py3.7-gcc7-no-ops	`ciflow/all`, `ciflow/cuda`, `ciflow/linux`, `ciflow/trunk`	🚫 skipped
macos-10-15-py3-arm64	`ciflow/all`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
macos-10-15-py3-lite-interpreter-x86-64	`ciflow/all`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
macos-11-py3-x86-64	`ciflow/all`, `ciflow/macos`, `ciflow/trunk`	🚫 skipped
parallelnative-linux-xenial-py3.7-gcc5.4	`ciflow/all`, `ciflow/cpu`, `ciflow/linux`, `ciflow/trunk`	🚫 skipped
periodic-libtorch-linux-bionic-cuda11.5-py3.7-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/libtorch`, `ciflow/linux`, `ciflow/scheduled`	🚫 skipped
periodic-libtorch-linux-xenial-cuda11.1-py3.7-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/libtorch`, `ciflow/linux`, `ciflow/scheduled`	🚫 skipped
periodic-linux-bionic-cuda11.5-py3.7-gcc7	`ciflow/all`, `ciflow/cuda`, `ciflow/linux`, `ciflow/scheduled`	🚫 skipped
periodic-linux-xenial-cuda10.2-py3-gcc7-slow-gradcheck	`ciflow/all`, `ciflow/cuda`, `ciflow/linux`, `ciflow/scheduled`, `ciflow/slow`, `ciflow/slow-gradcheck`	🚫 skipped
periodic-linux-xenial-cuda11.1-py3.7-gcc7-debug	`ciflow/all`, `ciflow/cuda`, `ciflow/linux`, `ciflow/scheduled`	🚫 skipped
periodic-win-vs2019-cuda11.1-py3	`ciflow/all`, `ciflow/cuda`, `ciflow/scheduled`, `ciflow/win`	🚫 skipped
periodic-win-vs2019-cuda11.5-py3	`ciflow/all`, `ciflow/cuda`, `ciflow/scheduled`, `ciflow/win`	🚫 skipped
pytorch-linux-xenial-py3-clang5-android-ndk-r19c-build	`ciflow/all`, `ciflow/android`, `ciflow/cpu`, `ciflow/linux`, `ciflow/trunk`	🚫 skipped

facebook-github-bot · 2022-01-31T14:32:47Z

🔗 Helpful links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/72059
📄 Preview docs built from this PR
📄 Preview C++ docs built from this PR
🔧 Opt-in to CIFlow to control what jobs run on your PRs

💊 CI failures summary and remediations

As of commit 4651368 (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).

Please report bugs/suggestions to the (internal) Dr. CI Users group.

Click here to manually regenerate this comment.

neerajprad

Some tests are failing. Note that you should be able to run your test locally using:

pytest test/distributions/test_distributions.py -k {name of test}

You can build locally with pytest setup.py build. See this for details.

neerajprad · 2022-02-01T04:29:15Z


        # Implemented Sampling using Bartlett decomposition
-        noise = self._dist_chi2.rsample(sample_shape).sqrt().diag_embed(dim1=-2, dim2=-1)
+        noise = _clamp_with_eps(


I suppose this is the important change in this PR. Since there are some unrelated changes as well, could you add the specific change for better numerical stability in the PR description so that it is available for future reference?

Thank you for the feedback. That would be great. I will summarize the Key modifications in the PR description.

Note that Chi2 (inherits from Gamma) already clamps based on .tiny. See

pytorch/torch/distributions/gamma.py

Line 62 in a482aeb

value.detach().clamp_(min=torch.finfo(value.dtype).tiny) # do not record in autograd graph

. I suppose this is adding some jitter to the diagonal elements, but is this high enough (I have seen 1e-6 or 1e-8 for float/double)?

Also, can we detach like in the example above so that its not recorded in autograd?

I suppose this is adding some jitter to the diagonal elements, but is this high enough (I have seen 1e-6 or 1e-8 for float/double)?

Thank you for the detailed points. I understand that it is not a neat way.
As you know, if we calculate reciprocal of float32 value clamped with tiny=1.17549e-38, we get numbers upper bounded by the exponent e^38 in base 10. But we get the number upper bounded by exponent e^7 if we clamp with eps=1e-7. Thus, I thought we need to clamp the value again to provide more stable code. Higher value might be better. I just adapted the information of the torch.finfo to make implementation to be seemed more principled.

Also, can we detach like in the example above so that its not recorded in autograd?

I will detach it if you think it is needed. But, I don't understand the intuition. Doesn't it affect .rsample()? I will check it out, too.

You are right, this will be used downstream so we shouldn't detach! Sorry about my hasty and misleading comment.

Thank you for the explanation. I think you found a potential issue in torch.distributions.gamma.Gamma. I wonder why there is a .detach either! Maybe we can fix it.

nonconvexopt · 2022-02-01T05:53:10Z

You can build locally with pytest setup.py build. See this for details.

I will try to utilize my local environment. I appreciate your kindness.

facebook-github-bot · 2022-02-15T17:07:55Z

@neerajprad has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

nonconvexopt · 2022-02-15T23:48:50Z

Thanks for the fix and more robust testing.

It is my pleasure. Thank you for the review.

…72059) Summary: Maintanance of #70377 Multiple modifications of the merged initial implementation of Wishart distribution. cc neerajprad Key modifications: - `torch/distributions/wishart.py`: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the `eps` value paired to each `torch.dtype` - `test/distributions/test_distributions.py`: Test Wishart distribution implementation in numerically unstable zones, i.e `df` values are at `ndim - 1 < df < ndim` where `ndim` is the one dimenstion of the Wishart parameter & sample matrix. Pull Request resolved: #72059 Reviewed By: H-Huang Differential Revision: D34245091 Pulled By: neerajprad fbshipit-source-id: 1cd653c1d5c663346433e84fd0bbe2e590790908

github-actions · 2022-02-16T16:40:19Z

Hey @nonconvexopt.
You've committed this PR, but it does not have both a 'release notes: ...' and 'topics: ...' label. Please add one of each to the PR. The 'release notes: ...' label should represent the part of PyTorch that this PR changes (fx, autograd, distributed, etc) and the 'topics: ...' label should represent the kind of PR it is (not user facing, new feature, bug fix, perf improvement, etc). The list of valid labels can be found here for the 'release notes: ...' and here for the 'topics: ...'.
For changes that are 'topic: not user facing' there is no need for a release notes label.

malfet · 2022-02-16T21:38:30Z

Reverted this PR as it broke MacOS CI, see https://github.com/pytorch/pytorch/runs/5221143435?check_suite_focus=true

facebook-github-bot · 2022-02-16T21:40:15Z

This pull request has been reverted by 0b117a3. To re-land this change, please open another pull request, assignthe same reviewers, fix the CI failures that caused the revert and make sure that the failing CI runs on the PR by applying the proper ciflow label (e.g., ciflow/trunk).

neerajprad · 2022-02-16T21:44:07Z

Here's the test failure for convenience:

======================================================================
FAIL [0.132s]: test_wishart_log_prob (__main__.TestDistributions)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "distributions/test_distributions.py", line 2321, in test_wishart_log_prob
    self.assertEqual(0.0, (batched_prob - unbatched_prob).abs().max(), atol=1e-3, rtol=0)
  File "/usr/local/miniconda/envs/build/lib/python3.8/site-packages/torch/testing/_internal/common_utils.py", line 2121, in assertEqual
    assert_equal(
  File "/usr/local/miniconda/envs/build/lib/python3.8/site-packages/torch/testing/_comparison.py", line 1081, in assert_equal
    raise error_metas[0].to_error()
AssertionError: Scalars are not close!

@malfet: Are these tests run after merging? I see that the CI was green on this PR, and I want to better understand how to check for issues like this before I merge.

malfet · 2022-02-16T21:45:39Z

@neerajprad due to the limited capacity, MacOS tests are not run on PRs by default, one can opt in by adding ciflow/macos label

nonconvexopt · 2022-02-17T02:29:55Z

I will try to fix the issue.

…(#72059) Summary: Maintanance of pytorch/pytorch#70377 Multiple modifications of the merged initial implementation of Wishart distribution. cc neerajprad Key modifications: - `torch/distributions/wishart.py`: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the `eps` value paired to each `torch.dtype` - `test/distributions/test_distributions.py`: Test Wishart distribution implementation in numerically unstable zones, i.e `df` values are at `ndim - 1 < df < ndim` where `ndim` is the one dimenstion of the Wishart parameter & sample matrix. Pull Request resolved: pytorch/pytorch#72059 Reviewed By: H-Huang Differential Revision: D34245091 Pulled By: neerajprad fbshipit-source-id: 1cd653c1d5c663346433e84fd0bbe2e590790908 (cherry picked from commit ef1da3ba465247f5777c3c40a90b96955c4281d0)

facebook-github-bot · 2022-02-22T22:18:03Z

This pull request has been reverted by 0b117a3. To re-land this change, please open another pull request, assignthe same reviewers, fix the CI failures that caused the revert and make sure that the failing CI runs on the PR by applying the proper ciflow label (e.g., ciflow/trunk).

…72993) Summary: Maintanance of #70377 Multiple modifications of the merged initial implementation of Wishart distribution. Key modifications: * torch/distributions/wishart.py: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the eps value paired to each torch.dtype * test/distributions/test_distributions.py: Test Wishart distribution implementation in numerically unstable zones, i.e df values are at ndim - 1 < df < ndim where ndim is the one dimenstion of the Wishart parameter & sample matrix. Re-opened reverted PR #72059 cc neerajprad vadimkantorov Pull Request resolved: #72993 Reviewed By: samdow Differential Revision: D34853807 Pulled By: neerajprad fbshipit-source-id: eb62dca19bf8a934fdf59b4ffc58587447fe8378

…72993) Summary: Maintanance of #70377 Multiple modifications of the merged initial implementation of Wishart distribution. Key modifications: * torch/distributions/wishart.py: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the eps value paired to each torch.dtype * test/distributions/test_distributions.py: Test Wishart distribution implementation in numerically unstable zones, i.e df values are at ndim - 1 < df < ndim where ndim is the one dimenstion of the Wishart parameter & sample matrix. Re-opened reverted PR #72059 cc neerajprad vadimkantorov Pull Request resolved: #72993 Reviewed By: samdow Differential Revision: D34853807 Pulled By: neerajprad fbshipit-source-id: eb62dca19bf8a934fdf59b4ffc58587447fe8378 (cherry picked from commit 99240c0)

…ytorch#72059) Summary: Maintanance of pytorch#70377 Multiple modifications of the merged initial implementation of Wishart distribution. cc neerajprad Key modifications: - `torch/distributions/wishart.py`: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the `eps` value paired to each `torch.dtype` - `test/distributions/test_distributions.py`: Test Wishart distribution implementation in numerically unstable zones, i.e `df` values are at `ndim - 1 < df < ndim` where `ndim` is the one dimenstion of the Wishart parameter & sample matrix. Pull Request resolved: pytorch#72059 Reviewed By: H-Huang Differential Revision: D34245091 Pulled By: neerajprad fbshipit-source-id: 1cd653c1d5c663346433e84fd0bbe2e590790908 (cherry picked from commit ef1da3b)

…ytorch#72993) Summary: Maintanance of pytorch#70377 Multiple modifications of the merged initial implementation of Wishart distribution. Key modifications: * torch/distributions/wishart.py: Clamp (Clip) float type values to calculate reciprocal in numerically stable manner, by using the eps value paired to each torch.dtype * test/distributions/test_distributions.py: Test Wishart distribution implementation in numerically unstable zones, i.e df values are at ndim - 1 < df < ndim where ndim is the one dimenstion of the Wishart parameter & sample matrix. Re-opened reverted PR pytorch#72059 cc neerajprad vadimkantorov Pull Request resolved: pytorch#72993 Reviewed By: samdow Differential Revision: D34853807 Pulled By: neerajprad fbshipit-source-id: eb62dca19bf8a934fdf59b4ffc58587447fe8378 (cherry picked from commit 99240c0)

nonconvexopt added 3 commits January 31, 2022 23:14

Clean codes

edf8257

Improve numerical stability

bd5ac88

Check numerical stability more rigorously

ec43666

pytorch-bot Bot added the ciflow/default label Jan 31, 2022

facebook-github-bot added the cla signed label Jan 31, 2022

nonconvexopt commented Jan 31, 2022

View reviewed changes

Comment thread torch/distributions/wishart.py

pytorchbot added the open source label Jan 31, 2022

nonconvexopt added a commit to nonconvexopt/pytorch that referenced this pull request Jan 31, 2022

Seperate the changes in wishart to pytorch#72059

d22d960

nonconvexopt added 3 commits February 1, 2022 10:24

Fix typo

e533bb4

Set df values in test larger than (ndim - 1)

5fde921

Fix typo

4bf3f2c

neerajprad reviewed Feb 1, 2022

View reviewed changes

nonconvexopt added 2 commits February 1, 2022 13:40

Extend .entropy to consider multivariate natural parameters

2a7b806

Increase tolerance value for .assertEqual

5a18141

ngimel added the triaged This issue has been looked at a team member, and triaged and prioritized into an appropriate module label Feb 2, 2022

nonconvexopt added 12 commits February 3, 2022 10:10

Consider SciPy version and adjust test conditions

6614bb6

Merge branch 'master' into distributions/wishart_improvement_v1

8012a14

Remove trailing spaces

57f618b

Fix typo and debug error in version comparison

aa71dd1

reduce variance in MC estimate

ade2dba

Check correct scipy version allowing df < ndim for Wishart

2147975

Allow higher error in MC estimate for moment

8af89d9

Apply harder clamping

934a626

Remove unnecessary code pieces

ba5b413

Remove experimental changes

3591ea6

Test numerical instability only in relevant tests

fc4faf7

Merge branch 'master' into distributions/wishart_improvement_v1

0dda7ca

pytorchmergebot closed this in 5343cfe Feb 16, 2022

facebook-github-bot added the Reverted label Feb 16, 2022

nonconvexopt mentioned this pull request Feb 17, 2022

Improve numerical stability of torch.distributions.wishart.Wishart #72993

Closed

janeyx99 mentioned this pull request Feb 18, 2022

[Meta] CI Revert Tracker #66178

Closed

Conversation

nonconvexopt commented Jan 31, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot Bot commented Jan 31, 2022

⚛️ CI Flow

Uh oh!

facebook-github-bot commented Jan 31, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful links

💊 CI failures summary and remediations

Uh oh!

Uh oh!

neerajprad left a comment

Choose a reason for hiding this comment

Uh oh!

neerajprad Feb 1, 2022

Choose a reason for hiding this comment

Uh oh!

nonconvexopt Feb 1, 2022

Choose a reason for hiding this comment

Uh oh!

neerajprad Feb 15, 2022

Choose a reason for hiding this comment

Uh oh!

nonconvexopt Feb 15, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nonconvexopt Feb 15, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

neerajprad Feb 15, 2022

Choose a reason for hiding this comment

Uh oh!

nonconvexopt Feb 16, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nonconvexopt commented Feb 1, 2022

Uh oh!

facebook-github-bot commented Feb 15, 2022

Uh oh!

nonconvexopt commented Feb 15, 2022

Uh oh!

github-actions Bot commented Feb 16, 2022

Uh oh!

malfet commented Feb 16, 2022

Uh oh!

facebook-github-bot commented Feb 16, 2022

Uh oh!

neerajprad commented Feb 16, 2022

Uh oh!

malfet commented Feb 16, 2022

Uh oh!

nonconvexopt commented Feb 17, 2022

Uh oh!

facebook-github-bot commented Feb 22, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

nonconvexopt commented Jan 31, 2022 •

edited

Loading

facebook-github-bot commented Jan 31, 2022 •

edited

Loading

nonconvexopt Feb 15, 2022 •

edited

Loading

nonconvexopt Feb 15, 2022 •

edited

Loading

nonconvexopt Feb 16, 2022 •

edited

Loading