[GPU] RMSFusion: support Multiply(x,x) squaring and optional Reshape#36120
Open
daijh wants to merge 1 commit into
Open
[GPU] RMSFusion: support Multiply(x,x) squaring and optional Reshape#36120daijh wants to merge 1 commit into
daijh wants to merge 1 commit into
Conversation
The TFLite RMS norm decomposition (odml.rms_norm) uses Multiply(x, x) instead of Power(x, 2) for squaring, and may insert a Reshape between the Add(mean, eps) and the rsqrt Power node when keepdims=false. Extend the RMSFusion pattern to handle: - Multiply(x, x) as an alternative to Power(x, 2) for squaring - Optional Reshape between add_eps and sqrt/rsqrt nodes Add corresponding unit tests for the new patterns.
github-actions
Bot
added
the
category: transformations
Author
|
This PR is ready for review, thanks. |
Author
|
Manual verify litert-lm gemma-3n-E2B on OpenVINO-GPU. |
Contributor
|
build_jenkins |
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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Details:
The TFLite RMS norm decomposition (odml.rms_norm) uses Multiply(x, x) instead of Power(x, 2) for squaring, and may insert a Reshape between the Add(mean, eps) and the rsqrt Power node when keepdims=false.
Extend the RMSFusion pattern to handle:
Add corresponding unit tests for the new patterns.
Tickets:
AI Assistance: