[QNN EP] Add Static bias encoding support for Conv op #26710
+572
−1
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
- Fix QNN-EP to support static bias tensors with mismatched quantization encodings - Add requantization logic to ensure bias_scale = weights_scale x activation_scale
Contributor
Author
|
Could we start the CI job for this PR ?@chilo-ms |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
The QNN-EP currently lacks support for static bias tensors with quantization encodings that don't match the expected mathematical relationship bias_scale = weights_scale × activation_scale.
Motivation and Context
This leads to HTP backend disregard the encodings of the bias tensor and recompute its own causing accuracy drops