torch.mtia module for MTIA device backend #123612

egienvalue · 2024-04-09T01:19:29Z

Stack from ghstack (oldest at bottom):

MTIA device has its own Module in PyTorch now.
torch.mtia has following APIs similar to other backends. The lazy_init is also supported.

__all__ = [
    "init",
    "is_available",
    "synchronize",
    "device_count",
    "current_device",
    "current_stream",
    "default_stream",
    "set_stream",
    "stream",
    "device",
]

For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon.

def _accelerator_hooks_device_count() -> _int: ...
def _accelerator_hooks_set_current_device(device_index: _int) -> None: ...
def _accelerator_hooks_get_current_device() -> _int : ...
def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ...
def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ...

Adding get_device_module API to retrieve device modules for different device types.

def get_device_module(device: Optional[Union[torch.device, str]] = None)

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- ghstack-source-id: 221767947 @exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

pytorch-bot · 2024-04-09T01:19:32Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/123612

📄 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 6571358 with merge base c5fafe9 ():
💚 Looks good so far! There are no failures yet. 💚

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

facebook-github-bot · 2024-04-09T01:19:47Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-09T01:53:55Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-09T05:51:50Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-09T05:53:29Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-09T15:27:55Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-09T18:23:25Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-09T18:41:52Z

This pull request was exported from Phabricator. Differential Revision: D52923602

albanD

The accelerator_hooks API looks a bit weird in this PR, but I think they make sense with the idea that we can have a base accelerator device module that uses them that is shared in the future.

aten/src/ATen/DeviceAccelerator.cpp

aten/src/ATen/detail/AcceleratorHooksInterface.h

docs/source/mtia.rst

torch/__init__.py

torch/mtia/__init__.py

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-10T03:43:37Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-10T05:36:21Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-10T20:06:08Z

This pull request was exported from Phabricator. Differential Revision: D52923602

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: torch.device) -> module: ``` --------- exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) [ghstack-poisoned]

facebook-github-bot · 2024-04-11T03:35:13Z

This pull request was exported from Phabricator. Differential Revision: D52923602

albanD

Thanks for adding this back. Not sure what is the right way to trigger all the necessary rocm signal. Please wait for @jeffdaily input here before merging.

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: Optional[Union[torch.device, str]] = None) ``` --------- [ghstack-poisoned]

jithunnair-amd · 2024-04-25T19:09:26Z

Thanks for adding this back. Not sure what is the right way to trigger all the necessary rocm signal. Please wait for @jeffdaily input here before merging.

Yes ciflow/rocm should be the extensive list of single-GPU unit tests for ROCm. Add ciflow/inductor (usually added automatically) for inductor-related PRs, and ciflow/periodic for distributed-logic related PRs.

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: Optional[Union[torch.device, str]] = None) ``` --------- Differential Revision: [D56443356](https://our.internmc.facebook.com/intern/diff/D56443356) Pull Request resolved: pytorch#123612 Approved by: https://github.com/albanD ghstack dependencies: pytorch#123611

…ytorch#123614) Test the generic torch.Stream/Event with fake device gurad and hooks. Differential Revision: [D56443358](https://our.internmc.facebook.com/intern/diff/D56443358) Pull Request resolved: pytorch#123614 Approved by: https://github.com/albanD ghstack dependencies: pytorch#123611, pytorch#123612

jeffdaily · 2024-04-26T16:00:38Z

@egienvalue @albanD There are some rocm jobs that run as part of ciflow/trunk before merge. A subset of the full ciflow/rocm suite. We don't have runner capacity to run all CI jobs on all PRs so we try to limit adding ciflow/rocm to all PRs, only the ones that are known to specifically target ROCm features or code paths. When a PR doesn't fail the subset of ROCm tests, but breaks after merge (unfortunate, yes), we recommend adding ciflow/rocm to the reverted PR at that point.

egienvalue · 2024-04-26T16:12:06Z

@egienvalue @albanD There are some rocm jobs that run as part of ciflow/trunk before merge. A subset of the full ciflow/rocm suite. We don't have runner capacity to run all CI jobs on all PRs so we try to limit adding ciflow/rocm to all PRs, only the ones that are known to specifically target ROCm features or code paths. When a PR doesn't fail the subset of ROCm tests, but breaks after merge (unfortunate, yes), we recommend adding ciflow/rocm to the reverted PR at that point.

@jeffdaily Thanks for the suggestion. I added that yesterday, all signals look good. I will try re-land them.

…123614) Test the generic torch.Stream/Event with fake device gurad and hooks. Since we added a fake device backend, it is mutual exclusive to other backends. Tests will be skipped if TEST_CUDA or TEST_ROCM is true. Pull Request resolved: #123614 Approved by: https://github.com/albanD ghstack dependencies: #123611, #123612

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: Optional[Union[torch.device, str]] = None) ``` --------- @exported-using-ghexport Differential Revision: [D52923602](https://our.internmc.facebook.com/intern/diff/D52923602/) Pull Request resolved: pytorch#123612 Approved by: https://github.com/albanD ghstack dependencies: pytorch#123611

…123614) Test the generic torch.Stream/Event with fake device gurad and hooks. @exported-using-ghexport Differential Revision: [D55902506](https://our.internmc.facebook.com/intern/diff/D55902506/) Pull Request resolved: #123614 Approved by: https://github.com/albanD ghstack dependencies: #123611, #123612

This reverts commit d7e1bf9. Reverted #123612 on behalf of https://github.com/jeffdaily due to This broke ROCm. see test_overrides.py ([comment](#123611 (comment)))

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: Optional[Union[torch.device, str]] = None) ``` --------- Differential Revision: [D56443356](https://our.internmc.facebook.com/intern/diff/D56443356) Pull Request resolved: #123612 Approved by: https://github.com/albanD ghstack dependencies: #123611

…123614) Test the generic torch.Stream/Event with fake device gurad and hooks. Differential Revision: [D56443358](https://our.internmc.facebook.com/intern/diff/D56443358) Pull Request resolved: #123614 Approved by: https://github.com/albanD ghstack dependencies: #123611, #123612

…ackend (pytorch#123614)" This reverts commit 355dc34. Reverted pytorch#123614 on behalf of https://github.com/jeffdaily due to this PR broke ROCm with message RuntimeError: Cannot have MTIA with other devices ([comment](pytorch#123612 (comment)))

This reverts commit 381653d. Reverted #123612 on behalf of https://github.com/jeffdaily due to this PR broke ROCm with message RuntimeError: Cannot have MTIA with other devices ([comment](#123612 (comment)))

MTIA device has its own Module in PyTorch now. torch.mtia has following APIs similar to other backends. The lazy_init is also supported. ``` __all__ = [ "init", "is_available", "synchronize", "device_count", "current_device", "current_stream", "default_stream", "set_stream", "stream", "device", ] ``` ------------ For device management. We expand AccleratorHooksInterface to support generic device management and it can be used in both C++ and PyThon. ``` def _accelerator_hooks_device_count() -> _int: ... def _accelerator_hooks_set_current_device(device_index: _int) -> None: ... def _accelerator_hooks_get_current_device() -> _int : ... def _accelerator_hooks_exchange_device(device_index: _int) -> _int : ... def _accelerator_hooks_maybe_exchange_device(device_index: _int) -> _int : ... ``` --------- Adding get_device_module API to retrieve device modules for different device types. ``` def get_device_module(device: Optional[Union[torch.device, str]] = None) ``` --------- Pull Request resolved: #123612 Approved by: https://github.com/albanD ghstack dependencies: #123611

…ytorch#123614) Test the generic torch.Stream/Event with fake device gurad and hooks. Since we added a fake device backend, it is mutual exclusive to other backends. Tests will be skipped if TEST_CUDA or TEST_ROCM is true. Pull Request resolved: pytorch#123614 Approved by: https://github.com/albanD ghstack dependencies: pytorch#123611, pytorch#123612

facebook-github-bot added the fb-exported label Apr 9, 2024

This was referenced Apr 9, 2024

Build device generic torch.Stream and torch.Event based on c10::Stream/Event #123611

Closed

Add test_cpp_extensions tests for stream_and_event and mita_backend #123614

Closed

egienvalue requested a review from albanD April 9, 2024 15:31

albanD reviewed Apr 9, 2024

View reviewed changes

egienvalue mentioned this pull request Apr 10, 2024

Replace _StreamBase/_EventBase with torch.Stream/Event #123707

Closed

egienvalue requested a review from albanD April 11, 2024 16:12

albanD added the ciflow/rocm Trigger "default" config CI on ROCm label Apr 25, 2024

albanD approved these changes Apr 25, 2024

View reviewed changes

pytorchmergebot closed this in 73744a2 Apr 26, 2024

github-actions bot deleted the gh/egienvalue/5/head branch June 4, 2024 01:54

torch.mtia module for MTIA device backend #123612

torch.mtia module for MTIA device backend #123612

Uh oh!

Conversation

egienvalue commented Apr 9, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pytorch-bot bot commented Apr 9, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/123612

✅ No Failures

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

facebook-github-bot commented Apr 9, 2024

Uh oh!

albanD left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

facebook-github-bot commented Apr 10, 2024

Uh oh!

facebook-github-bot commented Apr 10, 2024

Uh oh!

facebook-github-bot commented Apr 10, 2024

Uh oh!

facebook-github-bot commented Apr 11, 2024

Uh oh!

albanD left a comment

Choose a reason for hiding this comment

Uh oh!

jithunnair-amd commented Apr 25, 2024

Uh oh!

jeffdaily commented Apr 26, 2024

Uh oh!

egienvalue commented Apr 26, 2024

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

egienvalue commented Apr 9, 2024 •

edited

Loading

pytorch-bot bot commented Apr 9, 2024 •

edited

Loading