[shoumikhin]

#24979 sized the sliding-window ring buffer from the largest prefill chunk, which defaults to the window size. In test_cuda_sequence_export_inputs the window is 16, so the ring now has 31 slots. But the reference model takes all 32 tokens in one write. The write wraps around the ring, so the reference logits come out wrong and the test fails with a max difference of about 0.33 (the limit is 0.05). This turned unittest / linux, unittest-editable / linux and unittest-release / linux red on main.

This sets the reference model's max prefill chunk to 32, so its ring holds the whole write. The exported program is unchanged.

Tested: ran the same model setup on CPU. The one-shot 32-token reference now matches the same tokens fed as two 16-token chunks exactly. Without the change it differs by 0.3335, the same number CI reports.

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/25038

• 📄 Preview Python docs built from this PR

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

❌ You can merge normally! (2 Unrelated Failures), 1 Unclassified Failure

As of commit b640c0d with merge base 311d96d ():

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 CUDA Builds / test-muse-glimmer-cuda-e2e-kquant-17gb-dflash-image-batching / linux-job (gh) (this job did not run on the merge base, so DrCI cannot tell whether the failure is pre-existing)
  ##[error]received SIGINT

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

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

• Test CPU Backend / test-cpu / test-backend-linux (cpu_fp32, models) / linux-job (gh) (trunk failure)
  /__w/executorch/executorch/pytorch/executorch/backends/cpu/runtime/providers/xnnpack/XNNPACKProvider.cpp:1128:16: error: call of overloaded ‘clamp_bound(const ptn::Argument&, const float&, float&)’ is ambiguous
• Test CPU Backend / test-cpu / test-backend-linux (cpu_fp32, operators) / linux-job (gh) (trunk failure)
  /__w/executorch/executorch/pytorch/executorch/backends/cpu/runtime/providers/xnnpack/XNNPACKProvider.cpp:1128:16: error: call of overloaded ‘clamp_bound(const ptn::Argument&, const float&, float&)’ is ambiguous

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

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.