Arm backend: Reduce Qwen3 VL E2E footprint

[bdemirb]

Use a compact synthetic Qwen3 VL config for full-model E2E tests. Keep real checkpoint shapes in layer tests, where memory is bounded. This avoids instantiating the full 2B text stack in nightly CI.

Signed-off-by: Baris Demir baris.demir@arm.com

Change-Id: I2d53cb76090b564c487558f8e88706ab0f01ed81

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/20328

• 📄 Preview Python docs built from this PR

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

❌ 1 New Failure, 25 Pending, 3 Unrelated Failures

As of commit 8944bcb with merge base c37b6c3 ():

NEW FAILURE - The following job has failed:

• Cadence Build & Test / vision-build / vision (gh)
  Input required and not supplied: aws-region

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• Cadence Build & Test / hifi-build / hifi4 (gh) (detected as infra flaky with no log or failing log classifier)

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

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

• pull / test-llama-runner-qnn-linux (fp32, qnn_16a16w, qnn) / linux-job (gh) (trunk failure)
• trunk / test-llama-runner-qnn-linux (fp32, qnn_16a16w, qnn) / linux-job (gh) (trunk failure)

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

[bdemirb]

This is a forward fix for the Qwen3 VL E2E coverage added in #20274.

The original E2E test instantiated the full Qwen3 VL text config
(28 layers, hidden size 2048, vocab size 151936), which can put too
much memory pressure on nightly CI during export/lowering.

This PR keeps the same E2E language/vision topology coverage across
TOSA/VGF and FP32/BF16, but switches the E2E test to a compact
synthetic config. Real checkpoint shapes remain covered by the layer
tests, where memory is bounded.

[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.