[SOLVED] Retrieve a list of model-specific POS tags using spaCy

Retrieve a list of model-specific POS tags using spaCy

I am looking for a way to get a list of all possibly usable POS tags for a specific language model in spaCy.

In an answer to another question, spaCy's TAG_MAP has been referenced to, but I am not sure how to access this. The documentation of spaCy says that this attribute has been replaced. Since spaCy only uses a specific subset of all POS tags for a specific language, I would like to retrieve a list of all POS tags that are currently used with the initialized language model.

I did currently just set up a model this way:

import spacy

tagger = spacy.load("de_dep_news_trf")

# TODO print(pos_tags)

Now, how do I print a list of all possible pos tags for this model?

Solution

For pretrained pipelines, you can check the labels on the model page, under the "Label Scheme" entry.

If your pipeline has a tagger, like the German one does, you can do this:

tagger = nlp.get_pipe("tagger")
print(tagger.labels)