Qaulcomm AI Engine Direct - Fix quantization annotation for per channel quant

[shewu-quic]

summary:

• Fix the 8a8w config in custom annotation
• Enable to set act observer and symmetric argument for per channel quant
• Remove unuse custom annotation in llama.py

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/7026

• 📄 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:

• [DomainsOnly] Jobs fail with GLIBC version not found

✅ No Failures

As of commit 5d85cf1 with merge base 97b58bb ():
💚 Looks good so far! There are no failures yet. 💚

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

[shewu-quic]

Hi @cccclai,

The PR is to enable to set the act_observer for per-channel quant which to align the observer with other nodes.
Could you help to take a look?

Thanks,
Hutton

[cccclai]

@navsud can you help with verifying

[navsud]

LGTM

Thanks for the update... Without this, the act observers were not consistent across the graph, especially when we want to use MinMaxObserver for 16-bit activations.

[cccclai]

Thanks!

[facebook-github-bot]

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