Do not delegate CoreML reflect/replicate pads beyond the last two dims

[hemanth1999k]

Summary

Core ML only supports reflect and replicate padding on the last two dimensions. ReflectionPad3d, ReplicationPad3d, or a reflect pad of only the depth dimension of a 5D input currently lower to Core ML without complaint, then fail when the program loads:

Padding for more than two dimensions only supports `constant` mode

CoreMLPartitioner now overrides support for these pads (aten.pad in reflect/replicate mode with a nonzero pad outside the last two dims, plus reflection_pad3d / replication_pad3d). ops_to_not_decompose therefore stops keeping them whole, they decompose into ops Core ML runs, and the model stays delegated and matches eager. 1D/2D reflect and replicate pads and constant pads are unchanged.

Fixes #11708
Fixes #11709

Test plan

Added test_nonconstant_pad_beyond_last_two_dims_is_decomposed to backends/apple/coreml/test/test_coreml_partitioner.py. It covers ReflectionPad3d, ReplicationPad3d, and a depth-only reflect pad. For each it checks that aten.pad.default is not in ops_to_not_decompose, lowers with to_edge_transform_and_lower, runs the program through executorch.runtime.Runtime, and compares with eager (atol/rtol 1e-2 because Core ML computes in fp16). It also checks that a 2D reflect pad is still kept whole.

python -m pytest backends/apple/coreml/test/test_coreml_partitioner.py

13/13 pass with the change; on main only the new test fails. Run on macOS 26.2, executorch 1.5.1 / main, torch 2.14.0, coremltools 9.0. ufmt and flake8 are clean.

Circular 3D pads still raise a ValueError at lowering, as they do on main; that is a separate issue and out of scope here.

cc @nil-is-all

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/23687

• 📄 Preview Python docs built from this PR

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

⚠️ 16 Awaiting Approval

As of commit 5e92fb7 with merge base f2575be ():

AWAITING APPROVAL - The following workflows need approval before CI can run:

• Apple (gh)
• Build documentation (gh)
• Build Presets (gh)
• Cadence Build & Test (gh)
• Lint (gh)
• periodic (gh)
• pull (gh)
• Test ARM Backend (gh)
• Test CoreML Backend (gh)
• Test Cortex-M Backend (gh)
• Test NXP Backend (gh)
• Test QNN Backend (gh)
• Test QNN Windows MSVC build (gh)
• Test Vulkan Backend (gh)
• Test WebGPU Backend (gh)
• Test XNNPACK Backend (gh)

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

[meta-cla]

Hi @hemanth1999k!

Thank you for your pull request and welcome to our community.

Action Required

In order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you.

Process

In order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA.

Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with CLA signed. The tagging process may take up to 1 hour after signing. Please give it that time before contacting us about it.

If you have received this in error or have any questions, please contact us at cla@meta.com. Thanks!

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: hemanth1999k / name: hemanth1999k (5e92fb7)

[hemanth1999k]

@pytorchbot label "release notes: apple"

[meta-cla]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks!