Repository navigation
[Fix][Relax][Frontend][ONNX] Support Shape outputs as Gather indices - #20179
Merged
Merged
Conversation
The Relax ONNX frontend preserves Shape outputs as ShapeExpr values. Gather assumed its indices operand always had a tensor type and accessed its dtype, which raised an AttributeError when a Shape output was used as indices. Materialize ShapeExpr indices as int64 tensors before the existing dtype and negative-index handling. Add an execution test that compares the imported model against ONNX Runtime. Fixes apache#20176.
cchung100m
approved these changes
Aug 25, 2026
cchung100m
left a comment
cchung100m
left a comment
Contributor
There was a problem hiding this comment.
LGTM, thanks to @Gunse11er 😄
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The Relax ONNX frontend preserves Shape outputs as ShapeExpr values. Gather assumed its indices operand always had a tensor type and accessed its dtype, which raised an AttributeError when a Shape output was used as indices.
Materialize ShapeExpr indices as int64 tensors before the existing dtype and negative-index handling. Add an execution test that compares the imported model against ONNX Runtime.
Fixes #20176.