Can't find factory for 'relation_extractor' for language English (en)

[Gowdasuma]

Hi Team, am not able to load the model, showing factory for 'relation_extractor' for language English (en) issue.
i'm using Jupyter notebook.

import spacy
nlp = spacy.load("rel_component/training/model-best")
nlp

ValueError Traceback (most recent call last)
Cell In[2], line 2
1 import spacy
----> 2 nlp = spacy.load("rel_component/training/model-best")
3 nlp

File ~\AppData\Local\anaconda3\lib\site-packages\spacy_init_.py:51, in load(name, vocab, disable, enable, exclude, config)
27 def load(
28 name: Union[str, Path],
29 *,
(...)
34 config: Union[Dict[str, Any], Config] = util.SimpleFrozenDict(),
35 ) -> Language:
36 """Load a spaCy model from an installed package or a local path.
37
38 name (str): Package name or model path.
(...)
49 RETURNS (Language): The loaded nlp object.
50 """
---> 51 return util.load_model(
52 name,
53 vocab=vocab,
54 disable=disable,
55 enable=enable,
56 exclude=exclude,
57 config=config,
58 )

File ~\AppData\Local\anaconda3\lib\site-packages\spacy\util.py:467, in load_model(name, vocab, disable, enable, exclude, config)
465 return load_model_from_package(name, **kwargs) # type: ignore[arg-type]
466 if Path(name).exists(): # path to model data directory
--> 467 return load_model_from_path(Path(name), **kwargs) # type: ignore[arg-type]
468 elif hasattr(name, "exists"): # Path or Path-like to model data
469 return load_model_from_path(name, **kwargs) # type: ignore[arg-type]

File ~\AppData\Local\anaconda3\lib\site-packages\spacy\util.py:539, in load_model_from_path(model_path, meta, vocab, disable, enable, exclude, config)
537 overrides = dict_to_dot(config, for_overrides=True)
538 config = load_config(config_path, overrides=overrides)
--> 539 nlp = load_model_from_config(
540 config,
541 vocab=vocab,
542 disable=disable,
543 enable=enable,
544 exclude=exclude,
545 meta=meta,
546 )
547 return nlp.from_disk(model_path, exclude=exclude, overrides=overrides)

File ~\AppData\Local\anaconda3\lib\site-packages\spacy\util.py:587, in load_model_from_config(config, meta, vocab, disable, enable, exclude, auto_fill, validate)
584 # This will automatically handle all codes registered via the languages
585 # registry, including custom subclasses provided via entry points
586 lang_cls = get_lang_class(nlp_config["lang"])
--> 587 nlp = lang_cls.from_config(
588 config,
589 vocab=vocab,
590 disable=disable,
591 enable=enable,
592 exclude=exclude,
593 auto_fill=auto_fill,
594 validate=validate,
595 meta=meta,
596 )
597 return nlp

File ~\AppData\Local\anaconda3\lib\site-packages\spacy\language.py:1847, in Language.from_config(cls, config, vocab, disable, enable, exclude, meta, auto_fill, validate)
1844 factory = pipe_cfg.pop("factory")
1845 # The pipe name (key in the config) here is the unique name
1846 # of the component, not necessarily the factory
-> 1847 nlp.add_pipe(
1848 factory,
1849 name=pipe_name,
1850 config=pipe_cfg,
1851 validate=validate,
1852 raw_config=raw_config,
1853 )
1854 else:
1855 assert "source" in pipe_cfg

File ~\AppData\Local\anaconda3\lib\site-packages\spacy\language.py:814, in Language.add_pipe(self, factory_name, name, before, after, first, last, source, config, raw_config, validate)
810 pipe_component, factory_name = self.create_pipe_from_source(
811 factory_name, source, name=name
812 )
813 else:
--> 814 pipe_component = self.create_pipe(
815 factory_name,
816 name=name,
817 config=config,
818 raw_config=raw_config,
819 validate=validate,
820 )
821 pipe_index = self._get_pipe_index(before, after, first, last)
822 self._pipe_meta[name] = self.get_factory_meta(factory_name)

File ~\AppData\Local\anaconda3\lib\site-packages\spacy\language.py:683, in Language.create_pipe(self, factory_name, name, config, raw_config, validate)
675 if not self.has_factory(factory_name):
676 err = Errors.E002.format(
677 name=factory_name,
678 opts=", ".join(self.factory_names),
(...)
681 lang_code=self.lang,
682 )
--> 683 raise ValueError(err)
684 pipe_meta = self.get_factory_meta(factory_name)
685 # This is unideal, but the alternative would mean you always need to
686 # specify the full config settings, which is not really viable.

ValueError: [E002] Can't find factory for 'relation_extractor' for language English (en). This usually happens when spaCy calls nlp.create_pipe with a custom component name that's not registered on the current language class. If you're using a Transformer, make sure to install 'spacy-transformers'. If you're using a custom component, make sure you've added the decorator @Language.component (for function components) or @Language.factory (for class components).

Available factories: attribute_ruler, tok2vec, merge_noun_chunks, merge_entities, merge_subtokens, token_splitter, doc_cleaner, parser, beam_parser, lemmatizer, trainable_lemmatizer, entity_linker, entity_ruler, tagger, morphologizer, ner, beam_ner, senter, sentencizer, spancat, spancat_singlelabel, span_finder, future_entity_ruler, span_ruler, textcat, textcat_multilabel, en.lemmatizer

Your Environment

• Operating System: Windows
• Python Version Used: 3.10
• spaCy Version Used:3.6.1
  transformers : 4.20.1
  en-core-web-lg: 3.3.0
  en-core-web-trf : 3.6.1
  en-pipeline : 0.0.0

[svlandeg]

Hi!

The relation_extractor is not a built-in functionality in spaCy. Assuming you're using a custom implementation inspired by e.g. this REL tutorial, then you have to make sure that this definition is exposed to your working environment:

@Language.factory(
    "relation_extractor",
    ...
)
def make_relation_extractor(
    nlp: Language, name: str, model: Model, *, threshold: float
): 
...

You can do so by importing the relevant function at run-time, by using entry-points and/or packaging the model as explained here and here.

The final part of the error message makes this explicit: spaCy's internal configuration & catalogue system searches for the "relation_extractor" factory but can't find it because it's defined in code that it hasn't seen at the time when you're running this spacy.load function. So, you need to make sure it's available.

I know this can all be a bit daunting if you haven't worked with the config system much before. A good introduction is this video by Ines: https://www.youtube.com/watch?v=BWhh3r6W-qE

Hope this helps!

[Gowdasuma]

Thank you @svlandeg

I will check the reference point and let you know.