zejun
> I'm using a tiny test network that is just one linear layer. Using the updated build I still get: > > /usr/local/lib/python3.10/dist-packages/torchvision/io/image.py:13: UserWarning: Failed to load image Python extension:...
Hi, @aaronenyeshi @mruberry Could you help review the PR? 😃 We instantiate the UTs on more device backends and enable it for XPU. Thank you.
> @zejun-chen , please check CI failures. Yes. Under track.
Hi, @aaronenyeshi @EikanWang Could you help review again? For now, it passes the CI cases. Thank you. 😄
@pytorchbot merge
Hi, @aaronenyeshi Thank you for review. We have met CI failures in this PR for a while but no root cause. Here are the failed cases: ``` test_profiler.py::TestProfilerMultithreaded::test_source_multithreaded_basic_work_in_main_thread_True test_profiler.py::TestProfilerMultithreaded::test_source_multithreaded_basic_work_in_main_thread_True test_profiler.py::TestProfilerMultithreaded::test_source_multithreaded_basic_work_in_main_thread_True...
Test the failed cases with the code changes made by this PR in the PR: https://github.com/pytorch/pytorch/pull/135970 https://github.com/pytorch/pytorch/pull/135971
> Test the failed cases with the code changes made by this PR in the PR: #135970 #135971 From the dummy PR CI results, we find the following CI failures...
Hi, We restore enabling cases on XPU for the test file **test_profiler.py** because there are some unstable cases failure when moving them outside of test class. Here are they: `test_profiler.py::TestProfilerMultithreaded::test_source_multithreaded_basic_work_in_main_thread_True`...
It seems CI error: ``` 2024-10-14T08:53:14.3576418Z return self.fake_tensor_converter.from_real_tensor( 2024-10-14T08:53:14.3577579Z File "/opt/conda/envs/py_3.9/lib/python3.9/site-packages/torch/_subclasses/fake_tensor.py", line 375, in from_real_tensor 2024-10-14T08:53:14.3578564Z out = self.meta_converter( 2024-10-14T08:53:14.3579555Z File "/opt/conda/envs/py_3.9/lib/python3.9/site-packages/torch/_subclasses/meta_utils.py", line 1660, in __call__ 2024-10-14T08:53:14.3580457Z r = self.meta_tensor(...