[MPS] Add support for flatbuffer serialization > 4GB

[DenisVieriu97]

Add support for serializing tensor weights > 4GB.

This change mirrors #1542 and #1543 for the MPS delegate, which enables serialization of constant weight data outside the flatbuffer blob.

Final raw bytes packed in the PT file have the following format:

| MPS Header     |
| Flatbuffer     |
| Constant data  |

cc @cccclai, @larryliu0820, @kimishpatel

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/4574

• 📄 Preview Python docs built from this PR

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

❌ 2 New Failures

As of commit 57f5845 with merge base 1b092e9 ():

NEW FAILURES - The following jobs have failed:

• Apple / test-demo-ios / macos-job (gh)
  RuntimeError: Command bash /Users/runner/work/_temp/exec_script failed with exit code 65
• Apple / upload-frameworks-ios (gh)
  Credentials could not be loaded, please check your action inputs: Could not load credentials from any providers

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

[cccclai]

Looks great. Thank you for sending the change!

edit: trigger more tests as the end to end stories + llama is in the trunk job and we want to have it cover

[facebook-github-bot]

@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[DenisVieriu97]

edit: trigger more tests as the end to end stories + llama is in the trunk job and we want to have it cover

Thanks @cccclai . Please let me know if I should be triggering these tests, or is this done automatically?

[cccclai]

edit: trigger more tests as the end to end stories + llama is in the trunk job and we want to have it cover

Thanks @cccclai . Please let me know if I should be triggering these tests, or is this done automatically?

yeah I've triggered them, and looks clean.

[cccclai]

Could you add this change? The test is failing internally...

Suggested change

	from backends.apple.mps.serialization.mps_graph_schema import Buffer
	from executorch.backends.apple.mps.serialization.mps_graph_schema import Buffer

[kirklandsign]

@cccclai @DenisVieriu97 Do we know why it's not consistent? Make it fail on GitHub is better in this case

[cccclai]

I'm not exactly sure, maybe related to how it's installed

[cccclai]

Thanks! Still have a lint error...https://github.com/pytorch/executorch/actions/runs/10311496755/job/28545193000?pr=4574

[DenisVieriu97]

Thanks! Still have a lint error...https://github.com/pytorch/executorch/actions/runs/10311496755/job/28545193000?pr=4574

Updated!

[facebook-github-bot]

@cccclai has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.