Add float scalar input support (#18751)

[digantdesai]

Summary:

The pybindings code handled bool and int scalar inputs but was missing
support for float (Python float → C++ double). This caused failures when
running models like Addmm that take float alpha/beta parameters, throwing
'Unsupported python type '.

Added py::isinstancepy::float_ handling to convert Python floats to
EValue(double) in both the portable and XNNPACK input processing paths.

Reviewed By: 3l1

Differential Revision: D99845426

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/18751

• 📄 Preview Python docs built from this PR

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

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• [CI[B200] Smoke test encounters CUDA Unknown error for dgxb200-03 and dgxb200-04

✅ You can merge normally! (2 Unrelated Failures)

As of commit b3dce6d with merge base 75ba558 ():

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

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

• pull / unittest / windows / windows-job (gh) (trunk failure)
  ##[error]The operation was canceled.
• pull / unittest-editable / windows / windows-job (gh) (trunk failure)
  ##[error]The operation was canceled.

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

[github-actions]

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

Adds Python float scalar input support to the ExecuTorch Python bindings by converting float → double and wrapping it in EValue, addressing runtime failures for exported models that take floating-point scalar parameters (e.g., alpha/beta).

Changes:

• Convert Python float inputs to EValue(double) in PyModule::run_method input processing.
• Convert Python float inputs to EValue(double) in PyMethod::set_inputs input processing.

Comments suppressed due to low confidence (2)

extension/pybindings/pybindings.cpp:815

• The exception message says inputs must be a flat list of tensors, but this code path explicitly supports non-tensor inputs (None/bool/int/float). Consider updating the message to reflect the actually supported scalar types to avoid confusing users when they pass an unsupported input type.

      } else {
        throw std::runtime_error(
            "Unsupported python type " + type_str +
            ". Ensure that inputs are passed as a flat list of tensors.");

extension/pybindings/pybindings.cpp:1145

• The exception message says inputs must be a flat list of tensors, but this conversion logic supports non-tensor inputs (None/bool/int/float). Consider updating the message to mention the supported scalar types to make debugging unsupported inputs clearer.

      } else {
        throw std::runtime_error(
            "Unsupported python type " + type_str +
            ". Ensure that inputs are passed as a flat list of tensors.");

---

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

[meta-codesync]

@digantdesai has exported this pull request. If you are a Meta employee, you can view the originating Diff in D99845426.

[Copilot]

Pull request overview

Copilot reviewed 1 out of 1 changed files in this pull request and generated 1 comment.

Comments suppressed due to low confidence (2)

extension/pybindings/pybindings.cpp:815

• The fallback error message says to "Ensure that inputs are passed as a flat list of tensors", but this function now explicitly accepts non-tensor inputs (None/bool/int/float). Consider updating the message to reflect the actual supported scalar types (and that the list must be flat, but not tensor-only), so users aren’t misled when they pass an unsupported scalar/object.

        throw std::runtime_error(
            "Unsupported python type " + type_str +
            ". Ensure that inputs are passed as a flat list of tensors.");

extension/pybindings/pybindings.cpp:1145

• The fallback error message says to "Ensure that inputs are passed as a flat list of tensors", but this method now supports non-tensor scalar inputs (None/bool/int/float). Updating the message to mention the supported scalar types (and/or pointing to documentation) would make debugging unsupported inputs easier.

        throw std::runtime_error(
            "Unsupported python type " + type_str +
            ". Ensure that inputs are passed as a flat list of tensors.");

---

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

[Copilot]

Float scalar inputs are now converted to EValue(double), but there doesn’t appear to be pybindings coverage that exercises a program with a double/float scalar input (existing tests mainly use tensor-only inputs). Adding an end-to-end pybindings test that exports a module taking a float scalar input (e.g., alpha/beta in addmm, or a simple x * alpha) would prevent regressions across both ExecuTorchModule.run_method and ExecuTorchMethod.set_inputs paths.