[ONNX FE][Transformation] GroupQueryAttention refine work for NPU static shape design#34980
Draft
bopeng1234 wants to merge 2 commits intoopenvinotoolkit:masterfrom
Draft
[ONNX FE][Transformation] GroupQueryAttention refine work for NPU static shape design#34980bopeng1234 wants to merge 2 commits intoopenvinotoolkit:masterfrom
bopeng1234 wants to merge 2 commits intoopenvinotoolkit:masterfrom
Conversation
…id data is in front of the buffer as MLAS does
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
initial refine work for ONNX GQA, insert new kv into the past kv, valid data is in front of the buffer as MLAS does
Details:
The old logic for NPU static shape KV design is:
But, refer to MLAS implementation, for static shape KV, the valid data is in the front. onnxruntime gqa
So, this PR applies the changes accordingly.
ScatterUpdate the current KV into the buffer, keep valid data in the front as below shown.

Tickets:
AI Assistance: