Fix bug ORTModelForFeatureExtraction #2267
Merged
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
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
Co-authored-by: Ilyas Moutawwakil <57442720+IlyasMoutawwakil@users.noreply.github.com>
yxtay
reviewed
May 15, 2025
echarlaix
approved these changes
May 15, 2025
IlyasMoutawwakil
approved these changes
May 15, 2025
4 tasks
IlyasMoutawwakil
added a commit
that referenced
this pull request
May 15, 2025
* Fix bug ORTModelForFeatureExtraction * Update optimum/onnxruntime/modeling_ort.py Co-authored-by: Ilyas Moutawwakil <57442720+IlyasMoutawwakil@users.noreply.github.com> * Update optimum/onnxruntime/modeling_ort.py * return_dict support for ORTModelForxxx * Add _warn_on_unhandled_inputs helper method to ORTModel * Add test for the return_dict feature * fix style --------- Co-authored-by: Ilyas Moutawwakil <57442720+IlyasMoutawwakil@users.noreply.github.com>
|
Is this really fixing the underlying issue? With this PR, native Python dicts are returned. However, transformers uses ModelOutput, or? At least for me I have this issue now: When using a sentence-transformers ONNX model, inference fails at this point: https://github.com/UKPLab/sentence-transformers/blob/master/sentence_transformers/models/Transformer.py#L449-L450 |
|
oh wow ! so transformers doesn't return a dict when return_dict is True, and returns a tuple when it's False 😵 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
What does this PR do?
This PR updates the
ORTModelForFeatureExtraction'sforwardmethod to accept**kwargs, ensuring better compatibility with Hugging Face Transformers' conventions. In particular, it adds proper support for thereturn_dictargument: when explicitly passed, the output is returned as a plain dictionary instead of being wrapped in aModelOutput. Additionally, any unused keyword arguments are logged with a warning to help users identify unsupported inputs while maintaining backward compatibility.Fixes issue : #2266