Bugfix/fix wikidata train entity linker #4509
Conversation
|
Hi @ZhuoruLin, thanks for your pull request! 👍 It looks like you haven't filled in the spaCy Contributor Agreement (SCA) yet. The agrement ensures that we can use your contribution across the project. Once you've filled in the template, put it in the If you've already included the Contributor Agreement in your pull request above, you can ignore this message. |
svlandeg
added
bug
| @@ -77,6 +77,8 @@ def main( | |||
| if labels_discard: | |||
| labels_discard = [x.strip() for x in labels_discard.split(",")] | |||
| logger.info("Discarding {} NER types: {}".format(len(labels_discard), labels_discard)) | |||
| else: | |||
| labels_discard = [] | |||
|
|
|||
There was a problem hiding this comment.
This fix is definitely good to have here, thanks!
I'm actually surprised this throws an error. I'm assuming this happens because of this line in pipes.pyx ? I was assuming that self.cfg.get("labels_discard", []) would return [] if the key for labels_discard is passed with value None in cfg, but I guess that's not the case then?
There was a problem hiding this comment.
Hi @svlandeg . I think that's due to the fact that at line 107 the key labels_discard got passed into config already. Therefore the self.cfg.get("labels_discard", []) would return the value for existing key rather than the default.
I am a big fan of SpaCy's NEL! Thank you for all the nice works!
There was a problem hiding this comment.
Yep, you're right :-) Could you perhaps also fix the code in pipes.pyx itself? In case that gets called from a different script than the wikidata_train_entity_linker.py one?
We could rewrite that line to two lines:
to_discard = self.cfg.get("labels_discard", [])
if to_discard and ent.label_ in to_discard:
Which would make it work with both None as value as well as []. Or something such :-)
There was a problem hiding this comment.
Just made the change :).
There was a problem hiding this comment.
Awesome - I think this can be merged :-) Thanks again!
ZhuoruLin
force-pushed
the
bugfix/fix-wikidata-train-entity-linker
branch
from
e757250
to
c6305b0
Compare
Fix a NoneType iteration error in bin/wikidata_train_entity_linker.py
Description
The None default value for argument 'labels_discard' would cause a NoneType iteration error in the predict method of EntityLinker Pipeline. Passing in an empty list instead solved the problem.
Types of change
Bugfix
Checklist