ValueError: [E002] Can't find factory for 'relation_extractor' for language English (en)

[HarshadaPatil26]

I've come across the issue:
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, parser, beam_parser, entity_linker, ner, beam_ner, entity_ruler, lemmatizer, tagger, morphologizer, senter, sentencizer, textcat, textcat_multilabel, en.lemmatizer

I am referring the link for realtion extractor:
https://github.com/explosion/projects/tree/v3/tutorials/rel_component

[svlandeg]

Hi! Can you provide a little more background as to which command exactly you're running? Are you running the example project as is, or are you trying to integrate the code in a custom code base?

[HarshadaPatil26]

I am running the example project as is.
After training model if i try to load model,
nlp=spacy.load(("/myproject/rel_component/training/model-best")
Then getting below error.
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, parser, beam_parser, entity_linker, ner, beam_ner, entity_ruler, lemmatizer, tagger, morphologizer, senter, sentencizer, textcat, textcat_multilabel, en.lemmatizer

[svlandeg]

Ah, that makes sense. The code that is loading the model also needs to know about the new, custom factories that you used, so you'll have to import the relevant functions:

# make the factory work
from rel_pipe import make_relation_extractor, score_relations

# make the config work
from rel_model import create_relation_model, create_classification_layer, create_instances, 

[HarshadaPatil26]

I have imported the above functions.. still it showing same error.

[pshrutiii]

I've also been trying to use rel_component tutorial project from git as is. Though the model can be trained, I'm also not able to use it
nlp = spacy.load("./training/model-best")
the above command keeps breaking with the same error as in this discussion. I tried your suggestion and added one of the imports but it gave me a new set of errors.

When you open rel_pipe.py, you'll see it has a factory method

Not quite sure what i'm doing wrong. I would appreciate any help to load the model.

[svlandeg]

Hi @pshrutiii!

Your code imports relation_extractor, but that's not actually a function or variable defined in rel_pipe.py. It's the name given to the function make_relation_extractor, and this name should be loaded into the registry when you import the corresponding function. Can you try again with the original snippet I cited:

# make the factory work
from rel_pipe import make_relation_extractor, score_relations

# make the config work
from rel_model import create_relation_model, create_classification_layer, create_instances, 

and let me know whether it works, or what errors you're getting?

[pshrutiii]

For the registry to pick up, which file should i modify with these imports? The closest match to your imports suggestion was custom_functions.py and evaluate.py - Since only custom_functions.py was missing one of the imports, i tweaked that and hit: !python -m spacy project run all_gpu again when i try to load: nlp = spacy.load("./training/model-best") -- it gives me the same error