When converting a graph to a module that came from MlirXlaOpKernel force the use of _output_shapes. #63434
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Outdated suggestions cannot be applied.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.
When converting a graph to a module that came from MlirXlaOpKernel force the use of _output_shapes.
When an Op is to be lowered by MlirXlaOpKernel all the output tensor shapes have been thrown away (except in the attribute) - this hasn't been a problem for most kernels because they generate the output shape based on inputs or other features; however, some Ops (e.g. _XlaHostComputeMlir') do not and are broken by the go/tuba bridge changes. This change forces the graph to module conversion to use the _output_shapes attribute, if present, to reintroduce the Op's shape.