Expose executorch.__version__

[shoumikhin]

Summary

import executorch; executorch.__version__ currently raises AttributeError, even though torch.__version__, numpy.__version__, etc. all work. The version data already exists: setup.py generates executorch/version.py (with __version__ and git_version) from version.txt at build time, mirroring torch/version.py. But executorch has no top-level __init__.py to re-export it, so the generated module is never surfaced on the package. (executorch.version.__version__ works today; the conventional executorch.__version__ does not.)

This adds src/executorch/__init__.py that re-exports the generated version, with a graceful fallback so import executorch never fails in an unbuilt source tree:

• built wheel / editable install -> uses the generated executorch/version.py (exact build version + git hash);
• otherwise -> falls back to installed distribution metadata via importlib.metadata;
• otherwise -> 0.0.0+unknown (never raises).

version.txt stays the single source of truth; no version string is hardcoded, and setup.py is unchanged (it already generates version.py in this directory).

Note for reviewers

This promotes executorch from an implicit (PEP 420) namespace package to a regular package. For the single-wheel layout this is the conventional setup and matches torch/numpy, but please confirm no consumer relies on namespace-package merging of executorch.* across multiple roots.

Test plan

import executorch
print(executorch.__version__)   # e.g. 1.4.0a0+<githash>
print(executorch.git_version)

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/20271

• 📄 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:

• [ROCm] MI350 CI jobs will have longer queue times due to CI migration

❌ 11 New Failures, 1 Cancelled Job, 13 Pending, 1 Unclassified Failure

As of commit 22a82d0 with merge base 0378fc4 ():

NEW FAILURES - The following jobs have failed:

• Cadence Build & Test / hifi-build / hifi4 (gh)
  Input required and not supplied: aws-region
• Cadence Build & Test / vision-build / vision (gh)
  Input required and not supplied: aws-region
• pull / test-llama-runner-qnn-linux (fp32, qnn_16a16w, qnn) / linux-job (gh)
  RuntimeError: Command docker exec -t 944b6505bdcbbfa870f60c4f95d06451c5be3ac6f14af8aaa882f305b8c53b8c /exec failed with exit code 1
• pull / test-llama-runner-qnn-linux (fp32, qnn_8a8w, qnn) / linux-job (gh)
  RuntimeError: Command docker exec -t abaa95856dcee2ffd5634292618e896f99d8cc1fea5ccc6b968c1d1807baabb1 /exec failed with exit code 1
• pull / test-lora-linux / linux-job (gh)
  RuntimeError: Command docker exec -t 592ca410d4fc7d67e1bd0da84a6baf89a7b760f29e51aafeea987825c2b8b3b1 /exec failed with exit code 1
• pull / test-qnn-models-linux (dl3) / linux-job (gh)
  RuntimeError: Command docker exec -t c0a64b51a239161c77fabd9908df9a89fdb1e283628cf7c8dd6f837a47fab931 /exec failed with exit code 1
• pull / test-qnn-models-linux (mv2) / linux-job (gh)
  RuntimeError: Command docker exec -t f3cc60135b53397291e9d40fbf5137f370d0a784f3d6186949191f02cd24ecb0 /exec failed with exit code 1
• pull / test-qnn-models-linux (mv3) / linux-job (gh)
  RuntimeError: Command docker exec -t ceb0666a095eace466ee42d0e8c3cbaa918f5a4b50099400308e60f2d46c1adb /exec failed with exit code 1
• pull / test-sqnr-static-llm-qnn-linux (smollm2_135m) / linux-job (gh)
  RuntimeError: Command docker exec -t 8f85a7c95a48eb5aeaf710d534b3cf74d865ee27584cbcf8b3a1c466caf4c21b /exec failed with exit code 1
• pull / test-static-llama-qnn-linux (stories_110m) / linux-job (gh)
  RuntimeError: Command docker exec -t fa26aad5fe8e0c32ea3447eee97702134e22419b811f0f3754d6fdc52ccc5f9e /exec failed with exit code 1
• pull / test-static-llama-qnn-linux (stories_260k_bc) / linux-job (gh)
  RuntimeError: Command docker exec -t 9ededaf5b2948012a921d6323a9669ce1d6f0f7c3319ee92c4c1444587d54693 /exec failed with exit code 1

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

• Test WebGPU Backend (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)

CANCELLED JOB - The following job was cancelled. Please retry:

• periodic / gather-models (gh)

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 a top-level executorch Python package initializer to expose conventional version attributes (executorch.__version__, executorch.git_version) while keeping imports resilient in unbuilt source-tree scenarios.

Changes:

• Introduces src/executorch/__init__.py to re-export __version__ and git_version from the generated executorch/version.py.
• Implements a fallback to importlib.metadata (and finally 0.0.0+unknown) so import executorch does not fail when version.py isn’t present.

---

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

[shoumikhin]

Thanks for the review — addressed both:

• Import resilience: switched to a relative import and narrowed the catch to ModuleNotFoundError, so the fallback only triggers when version.py is genuinely absent (unbuilt source tree). A present-but-broken version.py now surfaces its error instead of being silently swallowed.
• Smoke test: added test/test_version.py asserting import executorch succeeds and exposes __version__ (non-empty str) and git_version, covering both the built and unbuilt-source paths.

[shoumikhin]

Closing this — CI shows the approach is incompatible with ExecuTorch's packaging model.

Adding a top-level src/executorch/__init__.py converts executorch from a PEP 420 implicit namespace package into a regular package, which stops Python from merging executorch.* across multiple roots. Components that are built/installed into a separate location and rely on that namespace merging then fail to import. The QNN and LoRA jobs break with:

import executorch.backends.qualcomm.python.PyQnnManagerAdaptor as PyQnnManager
ModuleNotFoundError: No module named 'executorch.backends.qualcomm.python'

This is the namespace-package risk noted in the PR description, now confirmed by CI (11 jobs, deterministic). Since executorch must remain a namespace package, there's no way to add a package __init__.py (which is required to set __version__) without this regression.

The underlying gap is real — the build already generates executorch/version.py (mirroring torch/version.py) but it isn't surfaced as executorch.__version__. A namespace-safe fix would need a different mechanism (e.g. maintainers deciding on a packaging approach that doesn't reintroduce a top-level __init__.py). Closing this one; happy to follow up if the team wants to pursue a namespace-compatible approach.