fix: ONNX FARMReader model conversion is broken #3211
Merged
+26
−13
Conversation
This file contains 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
vblagoje
added this to the
https://github.com/deepset-ai/haystack/milestone/8 milestone
| @@ -626,8 +626,8 @@ def convert_to_onnx( | |||
| https://huggingface.co/transformers/main_classes/model.html#transformers.PreTrainedModel.from_pretrained | |||
| :return: None. | |||
| """ | |||
| language_model_class = LanguageModel.get_language_model_class(model_name) | |||
| if language_model_class not in ["Bert", "Roberta", "XLMRoberta"]: | |||
| language_model = get_language_model(model_name) | |||
There was a problem hiding this comment.
You might use here this function:
There was a problem hiding this comment.
Will try, thanks Bogdan
|
@bogdankostic when I move the onnx test to be executed after test_retrieval (mdr+memory) then both tests pass. |
Create new onnx tests, add first test
|
@bogdankostic with your suggested change this PR started to pass 🚀 |
|
I added dedicated onnx test file so we can start testing this part of the code base even better in the future |
bogdankostic
approved these changes
Sep 26, 2022
test/nodes/test_retriever.py
Outdated
| @@ -1,4 +1,5 @@ | |||
| import logging | |||
| import os | |||
There was a problem hiding this comment.
I think this import is not used either.
test/nodes/test_retriever.py
Outdated
| @@ -11,6 +12,7 @@ | |||
| from elasticsearch import Elasticsearch | |||
|
|
|||
| from haystack.document_stores import WeaviateDocumentStore | |||
| from haystack.nodes import FARMReader | |||
test/nodes/test_reader.py
Outdated
| @@ -1,4 +1,7 @@ | |||
| import math | |||
| import os | |||
| import tempfile | |||
There was a problem hiding this comment.
I think this import is not used.
There was a problem hiding this comment.
Aha, that's because we don't check imports for tests, will fix - thanks @bogdankostic
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.
Related Issues
Proposed Changes:
During a recent language and tokenization refactoring, we broke FarmReader ONNX conversion support. A simple ONNX conversion of a QA model fails; we don't have a test case for it.
How did you test it?
Added a unit test for FarmReader ONNX conversion
Notes for the reviewer
Is there a more straightforward way to figure out the model type (aside from loading it using
get_language_model)? See PR changes.Checklist