How do you apply a SpaCy project in practice?

[metalaureate]

This has got to be such a stupid question but I cannot figure out how to load a package to actually use it once I have run a project workflow. I've read and re-read the documentation and just don't get it, sorry.

I ran the training pipeline on this project: https://github.com/explosion/projects/tree/v3/tutorials/ner_tweets (spacy project run all)

And I exported the package (spacy project run package). This creates a tar.

But I cannot load the package without errors.

OSError: [E053] Could not read config file from ner_tweets/packages/en_ner_tweets-1.0.0/config.cfg

nlp=spacy.load("./ner_tweets/packages/en_ner_tweets-1.0.0")

How do I actually add the model to the pipeline? 🙃

[ljvmiranda921]

Hi @metalaureate , the tar output from the package command creates an installable Python package. You need to install it first before loading:

pip install path/to/en_ner_tweets-1.0.0.tar.gz

Then you can load it into Python by passing the appropriate name in spacy.load. To sanity-check if your environment has installed the pipeline, you can run python -m spacy info. You can see more from the docs.

Since you ran the whole training process for ner_tweets, you can also load the model directly. Something like this:

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

[metalaureate]

D'oh