[Contrib][TRT] Fix Conv2D construction when channels attribute is not available. #6805
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Outdated suggestions cannot be applied.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.
This very small PR changes how channels are extracted from a Conv2D relay op when converting to TensorRT. Previously, we relied on the
channels
attribute being available. However, it is optional and often left asNone
. When attempting to convert in this casestoi
fails with an unpleasant error. Since we are already enforcing that the weight layout isOIHW
, there is no downside to instead extracting the channel info from the weight shape. I updated the corresponding conv2d test to not specify channels to confirm that this change resolves the issue.