Conversation
Contributor
|
@ochougul @quic-rishinr do we need this? |
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Gemma4 was the only QEff model still using HF’s runtime rotary embeddings, recomputing RoPE cos/sin via
Cos/Sinops on every forward pass. This caused the computations to be repeatedly traced into the exported ONNX graph across decoder and vision-encoder layers, triggering a compiler tensor-aliasing issue.The fix introduces
QEffGemma4TextRotaryEmbeddingandQEffGemma4VisionRotaryEmbedding, which precompute static cos/sin tables during initialization. As a result,forward()now performs a simpleGather, aligning Gemma4 with the implementation pattern used by all other QEff models and eliminating the decode corruption issue.