Dispatch take_along_axis to gather #107711

lezcano · 2023-08-22T16:56:37Z

Stack from ghstack (oldest at bottom):

Gather does the same thing, but it's much better supported in the
torch.compile stack

cc @mruberry @rgommers @voznesenskym @penguinwu @anijain2305 @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @chenyang78 @aakhundov

Gather does the same thing, but it's much better supported in the `torch.compile` stack [ghstack-poisoned]

pytorch-bot · 2023-08-22T16:56:40Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/107711

📄 Preview Python docs built from this PR
📄 Preview C++ docs built from this PR
❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit e8d25cd with merge base ba5eeed ():
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

Gather does the same thing, but it's much better supported in the `torch.compile` stack cc mruberry rgommers voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]

ezyang

tests?

Gather does the same thing, but it's much better supported in the `torch.compile` stack cc mruberry rgommers voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]

lezcano · 2023-08-22T22:24:24Z

torch/testing/_internal/common_methods_invocations.py

-    yield SampleInput(
-        make_arg(),
-        0,
-        torch.tensor([0], dtype=torch.int64, device=device))


Repeated sample.

@lezcano one of these is torch.tensor([0]) (i.e. 1d) and the other is torch.tensor(0) (i.e. 0d)

Ah, right, sorry for that. Reverting at #107776

lezcano · 2023-08-22T22:38:11Z

@pytorchbot merge

pytorchmergebot · 2023-08-22T22:39:48Z

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

pytorchmergebot · 2023-08-22T22:49:57Z

Merge failed

Reason: 1 mandatory check(s) failed. The first few are:

Lint / lintrunner / linux-job

Dig deeper by viewing the failures on hud

Details for Dev Infra team

Raised by workflow job

Failing merge rule: Core Maintainers

Gather does the same thing, but it's much better supported in the `torch.compile` stack cc mruberry rgommers voznesenskym penguinwu anijain2305 EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx chenyang78 aakhundov [ghstack-poisoned]

lezcano · 2023-08-22T22:55:54Z

@pytorchbot merge

pytorchmergebot · 2023-08-22T22:59:31Z

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here

As per title. Tests in the next PR Pull Request resolved: #107746 Approved by: https://github.com/ezyang ghstack dependencies: #107687, #107688, #107710, #107711

Fixes #107437 Pull Request resolved: #107689 Approved by: https://github.com/ezyang ghstack dependencies: #107687, #107688, #107710, #107711, #107746

ev-br · 2023-08-24T13:42:54Z

This seems to fail on main locally:

$ pytest test/torch_np/numpy_tests/lib/test_shape_base_.py
Test session starts (platform: linux, Python 3.8.17, pytest 7.4.0, pytest-sugar 0.9.7)
rootdir: /home/ev-br/repos/pytorch
configfile: pytest.ini
plugins: hypothesis-6.82.6, sugar-0.9.7, xdist-3.3.1
collected 72 items                                                                                                                                                                                        

 test/torch_np/numpy_tests/lib/test_shape_base_.py ✓✓✓                                                                                                                                        4% ▌         

―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――― TestTakeAlongAxis.test_broadcast ―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
Traceback (most recent call last):
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/_pytest/runner.py", line 341, in from_call
    result: Optional[TResult] = func()
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/_pytest/runner.py", line 262, in <lambda>
    lambda: ihook(item=item, **kwds), when=when, reraise=reraise
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_hooks.py", line 433, in __call__
    return self._hookexec(self.name, self._hookimpls, kwargs, firstresult)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_manager.py", line 112, in _hookexec
    return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_callers.py", line 155, in _multicall
    return outcome.get_result()
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_result.py", line 108, in get_result
    raise exc.with_traceback(exc.__traceback__)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_callers.py", line 80, in _multicall
    res = hook_impl.function(*args)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/_pytest/runner.py", line 177, in pytest_runtest_call
    raise e
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/_pytest/runner.py", line 169, in pytest_runtest_call
    item.runtest()
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/_pytest/python.py", line 1788, in runtest
    self.ihook.pytest_pyfunc_call(pyfuncitem=self)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_hooks.py", line 433, in __call__
    return self._hookexec(self.name, self._hookimpls, kwargs, firstresult)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_manager.py", line 112, in _hookexec
    return self._inner_hookexec(hook_name, methods, kwargs, firstresult)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_callers.py", line 116, in _multicall
    raise exception.with_traceback(exception.__traceback__)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/pluggy/_callers.py", line 80, in _multicall
    res = hook_impl.function(*args)
  File "/home/ev-br/.conda/envs/pytorch-dev/lib/python3.8/site-packages/_pytest/python.py", line 194, in pytest_pyfunc_call
    result = testfunction(**testargs)
  File "/home/ev-br/repos/pytorch/test/torch_np/numpy_tests/lib/test_shape_base_.py", line 101, in test_broadcast
    actual = take_along_axis(a, ai, axis=1)
  File "/home/ev-br/repos/pytorch/torch/_numpy/_normalizations.py", line 213, in wrapped
    result = func(*args, **kwds)
  File "/home/ev-br/repos/pytorch/torch/_numpy/_funcs_impl.py", line 889, in take_along_axis
    return torch.gather(arr, axis, indices)
RuntimeError: Size does not match at dimension 2 expected index [1, 2, 5] to be smaller than self [3, 4, 1] apart from dimension 1

 test/torch_np/numpy_tests/lib/test_shape_base_.py ⨯✓x✓✓xxxxxxxxx✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓✓sx✓✓✓✓✓✓✓✓✓✓✓x                                                                    100% ██████████
========================================================================================= short test summary info =========================================================================================
FAILED [0.0032s] test/torch_np/numpy_tests/lib/test_shape_base_.py::TestTakeAlongAxis::test_broadcast - RuntimeError: Size does not match at dimension 2 expected index [1, 2, 5] to be smaller than self [3, 4, 1] apart from dimension 1

Results (0.51s):
      58 passed
       1 failed
         - test/torch_np/numpy_tests/lib/test_shape_base_.py:97 TestTakeAlongAxis.test_broadcast
      12 xfailed
       1 skipped

Reverting the change from take_along_dim to torch.gather fixes the failure.

lezcano · 2023-08-24T14:13:26Z

It does! Let me put up a forward fix. This means that not all the tests in torch_np/ are run on CI. @albanD could you give us a hand here? What'd be the best way to add these tests to CI?

Addresses #107711 (comment) I'll have someone submit a decomposition of `take_along_dim` next week. [ghstack-poisoned]

Addresses #107711 (comment) I'll have someone submit a decomposition of `take_along_dim` next week. ghstack-source-id: f59e75463653db9086a519c46165afe1ce926136 Pull Request resolved: #107875

Dispatch take_along_axis to gather

9159f54

Gather does the same thing, but it's much better supported in the `torch.compile` stack [ghstack-poisoned]

lezcano mentioned this pull request Aug 22, 2023

Add a flag that allows breaking on NumPy ops #107687

Closed

This was referenced Aug 22, 2023

Support negative indices in ndarray.__getitem__ #107688

Closed

[dynamo] np.sort(complex) is not implemented #107710

Closed

[dynamo] Wrap ndarray dunder methods #107689

Closed

lezcano requested a review from ezyang August 22, 2023 17:05

lezcano added module: numpy Related to numpy support, and also numpy compatibility of our operators module: dynamo release notes: dynamo labels Aug 22, 2023

pytorchbot added the open source label Aug 22, 2023

lezcano requested a review from ev-br August 22, 2023 18:59

ezyang approved these changes Aug 22, 2023

View reviewed changes

lezcano mentioned this pull request Aug 22, 2023

[dynamo] Fix ndarray.__pow__ #107746

Closed

lezcano commented Aug 22, 2023

View reviewed changes

pytorch-bot bot added the ciflow/trunk Trigger trunk jobs on your pull request label Aug 22, 2023

pytorchmergebot added the merging label Aug 22, 2023

pytorchmergebot removed the merging label Aug 22, 2023

pytorchmergebot added the merging label Aug 22, 2023

pytorchmergebot added Merged and removed merging labels Aug 23, 2023

pytorchmergebot closed this in fada052 Aug 23, 2023

pytorchmergebot pushed a commit that referenced this pull request Aug 23, 2023

[dynamo] Fix ndarray.__pow__ (#107746)

b5c90ba

As per title. Tests in the next PR Pull Request resolved: #107746 Approved by: https://github.com/ezyang ghstack dependencies: #107687, #107688, #107710, #107711

pytorchmergebot pushed a commit that referenced this pull request Aug 23, 2023

[dynamo] Wrap ndarray dunder methods (#107689)

207b06d

Fixes #107437 Pull Request resolved: #107689 Approved by: https://github.com/ezyang ghstack dependencies: #107687, #107688, #107710, #107711, #107746

lezcano mentioned this pull request Aug 24, 2023

[dynamo] Dispatch take_along_axis to take_along_dim #107875

Closed

lezcano added a commit that referenced this pull request Aug 24, 2023

[dynamo] Dispatch take_along_axis to take_along_dim

daaa379

Addresses #107711 (comment) I'll have someone submit a decomposition of `take_along_dim` next week. [ghstack-poisoned]

facebook-github-bot deleted the gh/Lezcano/225/head branch August 26, 2023 14:16

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Dispatch take_along_axis to gather #107711

Dispatch take_along_axis to gather #107711

lezcano commented Aug 22, 2023 •

edited

pytorch-bot bot commented Aug 22, 2023 •

edited

ezyang left a comment

lezcano Aug 22, 2023

peterbell10 Aug 23, 2023

lezcano Aug 23, 2023

lezcano commented Aug 22, 2023

pytorchmergebot commented Aug 22, 2023

pytorchmergebot commented Aug 22, 2023

lezcano commented Aug 22, 2023

pytorchmergebot commented Aug 22, 2023

ev-br commented Aug 24, 2023

lezcano commented Aug 24, 2023

Dispatch take_along_axis to gather #107711

Dispatch take_along_axis to gather #107711

Conversation

lezcano commented Aug 22, 2023 • edited

pytorch-bot bot commented Aug 22, 2023 • edited

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/107711

✅ No Failures

ezyang left a comment

Choose a reason for hiding this comment

lezcano Aug 22, 2023

Choose a reason for hiding this comment

peterbell10 Aug 23, 2023

Choose a reason for hiding this comment

lezcano Aug 23, 2023

Choose a reason for hiding this comment

lezcano commented Aug 22, 2023

pytorchmergebot commented Aug 22, 2023

Merge started

pytorchmergebot commented Aug 22, 2023

Merge failed

lezcano commented Aug 22, 2023

pytorchmergebot commented Aug 22, 2023

Merge started

ev-br commented Aug 24, 2023

lezcano commented Aug 24, 2023

lezcano commented Aug 22, 2023 •

edited

pytorch-bot bot commented Aug 22, 2023 •

edited