[pytorchbot]

Merge after #24298. This pick is stacked on it, because #24400 builds on changes that #24298 brings to release/1.6.

to_edge cannot lower F.interpolate(x, mode="bicubic", antialias=True) today. It exports to aten._upsample_bicubic2d_aa, which is not a core ATen op, so to_edge stops with:

SpecViolationError: Operator torch._ops.aten._upsample_bicubic2d_aa.default is not in Core ATen opset

ExecuTorch also has no kernel for it.

What this change does

It adds a portable kernel for _upsample_bicubic2d_aa.out and lists the op as an exception in the edge verifier, as the antialiased bilinear resize does.

Like ATen's CPU kernel, it uses the Keys cubic filter (a = -0.5), widened when shrinking, resizes the width, then the height, and copies a dim that keeps its size. float16 and bfloat16, which ATen's CPU kernel lacks, are computed in float32 and rounded once. uint8 follows ATen's fixed-point path: 16-bit integer weights, and each pass rounds and clamps. Other integer types fail in the kernel, as they do in ATen.

Between passes it keeps only the rows one output row reads, so shrinking 1080x1920 to 224x224 needs about 95 KB of temporary memory (43 KB for uint8). It rejects scales F.interpolate never produces: too small for the input, or (uint8 only) with weights too large for 16 bits. Where ATen's generic CPU path is wrong (contiguous input, output width 1, changing height; PyTorch issue 200568), it differs from eager on purpose.

The shared upsample helper now ignores a non-positive scale, as ATen does; of the other upsample kernels, only antialiased bilinear can pass one. A stale llava comment is updated, and the Arm all-ops test marks the kernel as needing a recipe. Like several portable kernels, it has no experimental CPU backend route yet.

Testing

• New kernel tests check against ATen: shrinking, align_corners, scales, windows past the input, kept dims, both layouts, a batch, all float types, uint8 bit for bit, a dynamic output, running out of temporary memory, and error cases. Each case catches a bug that no other case catches.
• The to_edge test compares static and dynamic programs (both layouts) with eager in float32 and, bit for bit, in uint8. It fails without the kernel or the verifier exception.
• Outside the suite, a random sweep matched ATen's uint8 results exactly on x86 and Arm, on both its vectorized and generic paths; float32 within 7e-6 for unit-scale inputs, half types within one unit in the last place.
• With XNNPACK, the resize stays outside the delegate. On one laptop thread, shrinking a contiguous 1080x1920 input to 224x224 takes about 7 ms (12 ms channels last).

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

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24447

• 📄 Preview Python docs built from this PR

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

❌ 3 Pending, 2 Unrelated Failures, 6 Unclassified Failures

As of commit bff92dc with merge base 4db778e ():

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:

• MLX / test-mlx-dflash / test-mlx-dflash (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  executorch
• MLX / test-mlx-llm (google/gemma-4-E2B-it, gemma4-e2b, false, 4w, macos-15-xlarge) / test-mlx-llm-gemma4-e2b-4w (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  executorch
• MLX / test-mlx-llm (unsloth/gemma-3-1b-it, gemma3-1b, false, 4w, macos-14-xlarge) / test-mlx-llm-gemma3-1b-4w (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  executorch
• MLX / test-mlx-stories110m / test-mlx-stories110m (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  executorch
• MLX / test-mlx-supertonic / test-mlx-supertonic (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  executorch
• Test CUDA Builds / unittest-cuda-quantized-gemm (int4) / linux-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  ##[error]fatal: Fetched in submodule path 'extension/llm/tokenizers', but it did not contain 3fc1df71045f76841ee912b31f1f176067de2e16. Direct fetching of that commit failed.

BROKEN TRUNK - The following jobs failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / nxp-mcuxpresso-test / linux-job (gh) (trunk failure)
  RuntimeError: Command bash /__w/_temp/exec_script failed with exit code 1
• Test CoreML Backend / test-coreml / test-backend-macos (coreml, models) / macos-job (gh) (trunk failure)
  ##[error]The operation was canceled.

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