Qualcomm AI Engine Direct - sliding attention lookahead support

[haowhsu-quic]

Summary

• add lookahead decode to speed up prompt calibration
• sliding attention lookahead support

Test plan

python examples/qualcomm/oss_scripts/llama/llama.py --decoder_model gemma3-1b -b build-android/ -m SM8750 -s 5f396958 --prompt "Could you tell me about Facebook?" --max_seq_len 1024 --kv_updater smart_mask --prefill_ar_len 16 --model_mode lookahead --compile_only --tasks wikitext --limit 1 --model_mode lookahead --window 4 --ngram 3 --gcap 4

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/14412

• 📄 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

As of commit d71a2a5 with merge base c9f46e2 ():

NEW FAILURE - The following job has failed:

• pull / test-samsung-models-linux / linux-job (gh)
  RuntimeError: Command docker exec -t cf02d32106449d91475bb35d6e489039720f297a4acb6e48261b85fa270eb0cf /exec failed with exit code 1

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

[haowhsu-quic]

@pytorchbot label "release notes: qualcomm"

[haowhsu-quic]

Hi @cccclai, this PR might not need to be in 1.0 release but #14411 will be required. Thank you.

[facebook-github-bot]

@cccclai has imported this pull request. If you are a Meta employee, you can view this in D83071531.