Revert "Arm backend: Make composable_quantizer default" by digantdesai · Pull Request #20982 · pytorch/executorch

digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


digantdesai · 2026-07-16T17:00:39Z

Reverts #19758

cc @freddan80 @per @zingo @oscarandersson8218 @mansnils @Sebastian-Larsson @robell @rascani

This reverts commit 432353a.

pytorch-bot · 2026-07-16T17:00:44Z

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20982

📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 3 Unclassified Failures

As of commit 913eb73 with merge base 2ecfc10 ():

NEW FAILURE - The following job has failed:

pull / unittest-nxp-neutron / linux-job (gh)
RuntimeError: Command docker exec -t c23b87fef3a6aeda997296ee57eca2ea911abaf5f8d266ea1210bc20b7bbf34c /exec failed with exit code 1

UNCLASSIFIED FAILURES - DrCI could not classify the following jobs because the workflow did not run on the merge base. The failures may be pre-existing on trunk or introduced by this PR:

Build Windows Wheels / pytorch/executorch / build-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Process completed with exit code 1.
Build Windows Wheels / pytorch/executorch / upload / upload-wheel-py3_10-cpu (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
Unable to download artifact(s): Artifact not found for name: pytorch_executorch__3.10_cpu_x64
Check Labels (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

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

github-actions · 2026-07-16T17:01:34Z

This PR needs a `release notes:` label

If your change should be included in the release notes (i.e. would users of this library care about this change?), please use a label starting with release notes:. This helps us keep track and include your important work in the next release notes.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "release notes: none"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Copilot

Pull request overview

This PR reverts the earlier change that made the Arm backend’s composable quantizer the default, restoring the legacy/default quantizer as the default implementation while keeping the composable quantizer behind an opt-in flag.

Changes:

Switch use_composable_quantizer defaults back to False across Arm quantizers, update warnings/docs accordingly, and refresh the public API manifest signatures.
Update Arm tests and tutorial/docs to align with the reverted default (including removing a composable-specific while-loop test).
Adjust quantization behavior around to.dtype integer→float output qspec handling and update related tests/expectations.

Reviewed changes

Copilot reviewed 13 out of 13 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
examples/arm/quantizer_tutorial.ipynb	Marks the composable-quantizer tutorial as WIP/experimental and adds an explicit warning + issue link.
docs/source/backends/arm-vgf/arm-vgf-quantization.md	Updates VGF quantizer default flag and warning text to reflect composable being experimental/opt-in.
docs/source/backends/arm-ethos-u/arm-ethos-u-quantization.md	Updates Ethos-U quantizer default flag and warning text to reflect composable being experimental/opt-in.
backends/arm/test/ops/test_while.py	Removes composable-quantizer-specific while-loop test and related imports.
backends/arm/test/ops/test_transpose_conv2d.py	Updates config type usage to `QuantizationConfig` and adjusts imports accordingly.
backends/arm/test/ops/test_to_copy.py	Tweaks TOSA INT cast test setup and splits redundant-cast xfails between FP/INT.
backends/arm/test/models/test_torch_functions.py	Simplifies `nonzero` test wiring and removes the special-case quantizer IO override.
backends/arm/test/misc/test_shared_qspecs.py	Updates expected qspec mappings/counts to match the reverted quantizer behavior.
backends/arm/test/misc/test_quant_custom_meta.py	Removes a quantizer IO override that was tied to the previous default behavior.
backends/arm/quantizer/quantization_config.py	Broadens integer→float `to.dtype` fixed-output qspec behavior from float32-only to any floating-point output.
backends/arm/quantizer/arm_quantizer.py	Restores default `use_composable_quantizer=False`, updates warnings/log messaging, and tightens legacy vs composable property behavior.
backends/arm/public_api_manifests/api_manifest_running.toml	Updates public API signatures for EthosU/VGF quantizers to reflect the reverted default parameter value.
backends/arm/_passes/normalize_while_initial_args_pass.py	Changes placeholder meta propagation for cloned placeholders in while normalization.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

 )
 cloned_placeholders.append(clone)
- clone.meta = placeholder.meta.copy()
+ clone.meta = placeholder.meta


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+- **use_composable_quantizer (bool)**: Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


- and will be removed in two minor releases. See
- [issue #17701](https://github.com/pytorch/executorch/issues/17701)
- for details.
+ use_composable_quantizer (bool): Whether to use the composable quantizer implementation. See https://github.com/pytorch/executorch/issues/17701" for details.


Revert "Arm backend: Make composable_quantizer default (#19758)"
913eb73
This reverts commit 432353a.

CopilotAI review requested due to automatic review settings July 16, 2026 17:00

digantdesai requested a review from mergennachin as a code owner July 16, 2026 17:00

pytorch-botBot added the ci-no-td label Jul 16, 2026

meta-claBot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jul 16, 2026

digantdesai temporarily deployed to cadence July 16, 2026 17:00 — with GitHub Actions Inactive

github-actionsBot added ciflow/trunk module: arm Issues related to arm backend labels Jul 16, 2026

Copilotstarted reviewing on behalf of digantdesai July 16, 2026 17:01 View session

CopilotAI reviewed Jul 16, 2026
View reviewed changes

rascani approved these changes Jul 16, 2026
View reviewed changes

digantdesai temporarily deployed to cadence July 16, 2026 17:29 — with GitHub Actions Inactive

rascani merged commit b9af901 into mainJul 17, 2026
521 of 528 checks passed

rascani deleted the revert-19758-change-1253855 branch July 17, 2026 16:29

rascani mentioned this pull request Jul 17, 2026
Reland "Arm backend: Make composable_quantizer default" #21018
Merged