[abm-77]

Summary

ReplaceScalarWithTensorArgPass passes the scalar to full through float(), which rounds an integer above 2**53. With the Arm int64 extension, x == 2**53 + 1 then compares against 2**53. Integer scalars now go to full unchanged, and float scalars are unchanged.

Split out of #24128 at the Arm maintainers' request. Found while adding int64 comparisons in #24622.

Test Plan

Added test_replace_scalar_with_tensor.py: an integer scalar of 2**53 + 1 is carried exactly and a float scalar is unchanged. With #24622, x == 2**53 + 1 and x > 2**53 + 1 under TOSA-1.1+INT+int64 match the reference model.

This PR was authored with Claude Code.

cc @nil-is-all @YufengShi-dudu @Erik-Lundell

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/24625

• 📄 Preview Python docs built from this PR

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

⚠️ 19 Awaiting Approval

As of commit 523020b with merge base d1f597e ():

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

• Build Aarch64 Linux Native Wheels (gh)
• Build documentation (gh)
• Build Linux Native Wheels (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 CPU Backend (gh)
• Test NXP Backend (gh)
• Test QNN Backend (gh)
• Test QNN Windows MSVC build (gh)
• Test Vulkan Backend (gh)
• Test Vulkan Backend (specialized runners) (gh)
• Test WebGPU Backend (gh)
• Test XNNPACK Backend (gh)

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

[Copilot]

🟢 Approval recommended

No active review issues are recorded for the change to preserve integer scalar precision during tensor replacement.

No open issues 🎉

What changed in this PR

This PR preserves large integer scalars in the shared ReplaceScalarWithTensorArgPass, supporting exact int64 comparisons in the related Arm backend work.

Changes:

• Pass integer scalars directly to full, avoiding precision loss from conversion to float.
• Add regression tests for exact handling of 2**53 + 1 and unchanged float handling.

File	Description
backends/​transforms/​test/​test_replace_scalar_with_tensor.py	Tests integer precision and unchanged float scalar handling.
backends/​transforms/​replace_scalar_with_tensor.py	Preserves integer scalars when creating replacement tensors.

_{Review effort: Balanced}

[Copilot]

🟢 Approval recommended

No active review issues are recorded for the change to preserve large integer scalars in the scalar-to-tensor transform.

No open issues 🎉

_{Review effort: Balanced}

[abm-77]

@pytorchbot label "release notes: none"