Skip to content

Conversation

@oscarandersson8218
Copy link
Collaborator

@oscarandersson8218 oscarandersson8218 commented Nov 12, 2025

Use the Node meta 'custom' field to propagate information from quantizer to partitioner using a new ArmAnnotationInfo data class. This allows us to track quantized node reliably which is useful in order to track which nodes should 'fold' it's quantization parameter and which should be kept in fp when mixing integer and float in a sub-graph.

cc @freddan80 @per @zingo @digantdesai

@pytorch-bot
Copy link

pytorch-bot bot commented Nov 12, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/executorch/15778

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

✅ No Failures

As of commit c557356 with merge base 8c48a76 (image):
💚 Looks good so far! There are no failures yet. 💚

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

@meta-cla meta-cla bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Nov 12, 2025
@oscarandersson8218 oscarandersson8218 added partner: arm For backend delegation, kernels, demo, etc. from the 3rd-party partner, Arm ciflow/trunk release notes: none Do not include this in the release notes labels Nov 12, 2025
@meta-codesync
Copy link

meta-codesync bot commented Nov 13, 2025

@SS-JIA has imported this pull request. If you are a Meta employee, you can view this in D86915881.

@SS-JIA
Copy link
Contributor

SS-JIA commented Nov 14, 2025

@oscarandersson8218 still failing due to

E   ModuleNotFoundError: No module named 'executorch.backends.arm.common.annotation_meta'

Mind adding the buck file change from #15759?

per and others added 2 commits November 14, 2025 08:34
Use the Node meta 'custom' field to propagate information from quantizer
to partitioner using a new ArmAnnotationInfo data class. This  allows us
to track quantized node reliably which is useful in order to track which
nodes should 'fold' it's quantization parameter and which should be kept
in fp when mixing integer and float in a sub-graph.

Co-authored-by: Per Åstrand <per.astrand@arm.com>
Signed-off-by: Oscar Andersson <oscar.andersson@arm.com>
Change-Id: I398bf52e14d58fce56aa46ace74e45f45050c81b
Signed-off-by: Oscar Andersson <oscar.andersson@arm.com>
Change-Id: I84aa43f880e0ea7bbefb30bf7f31fc0e3b362e5a
@oscarandersson8218 oscarandersson8218 force-pushed the node_quant_metadata_serializable branch from e308913 to c557356 Compare November 14, 2025 07:36
@zingo
Copy link
Collaborator

zingo commented Nov 14, 2025

Hi @SS-JIA as this and the internal Meta PR is different now this needs a re-import to get OK to merge e.g. we now get:

Meta Internal-Only Changes Check — The Diff and Pull Request are not in sync!

Thanks for your help with this PR (and other)

@SS-JIA
Copy link
Contributor

SS-JIA commented Nov 14, 2025

Tests look good now - feel free to merge!

@zingo zingo merged commit 3af3385 into pytorch:main Nov 14, 2025
299 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

ciflow/trunk CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. partner: arm For backend delegation, kernels, demo, etc. from the 3rd-party partner, Arm release notes: none Do not include this in the release notes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants