Fix ComplexTypeMark loss in TupleUnpack, ListUnpack, squeeze, select, narrow, type_as operations #33098
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
… with complex tensor outputs
…npack conversion when model has shape inference errors in other operations
Remove unnecessary formatting and refactoring changes: - Revert blank line additions in complex_type_mark.cpp - Revert comment fix in complex_type_mark.cpp - Revert variable reordering in complex_type_mark.cpp - Revert code simplification in unsqueeze.cpp These changes were made during debugging but are not required for the CVS-176305 fix.
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Outdated suggestions cannot be applied.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.
Details:
This PR fixes shape mismatch error when converting Qwen-Image model
Root cause:
ComplexTypeMarkmarker was lost when complex tensors passed through various operations. OpenVINO usesComplexTypeMarkto track complex tensors (represented as float tensors with trailing dimension of size 2). When the marker is lost, subsequent operations incorrectly add extra dimensions for "alignment".Changes:
ComplexTypeMarkRemover transformation - removes ComplexTypeMark nodes from final graph for models with complex tensor outputs
PrimListUnpackReplacer reordering - moved to no_val block to execute before validation
prim::TupleUnpack - preserve ComplexTypeMark through tuple unpacking (new translator + replacer fix)
prim::ListUnpack - preserve ComplexTypeMark in else-branch for unresolved cases
aten::squeeze - unwrap-operate-rewrap pattern to preserve ComplexTypeMark
aten::select - unwrap-operate-rewrap pattern to preserve ComplexTypeMark
aten::narrow - unwrap-operate-rewrap pattern to preserve ComplexTypeMark
aten::type_as - dedicated translator using ComplexTypeMark::convert_like() instead of generic ConvertLike
Tickets: