Skip to content

๐Ÿ“‚ Additional lookup tables and data resources for spaCy

License

Notifications You must be signed in to change notification settings

explosion/spacy-lookups-data

Folders and files

NameName
Last commit message
Last commit date

Latest commit

ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 
ย 

Repository files navigation

spaCy lookups data

This repository contains additional data files to be used with spaCy v2.2+. When it's installed in the same environment as spaCy, this package makes the resources for each language available as an entry point, which spaCy checks when setting up the Vocab and Lookups.

Feel free to submit pull requests to update the data. For issues related to the data, lookups and integration, please use the spaCy issue tracker.

tests Current Release Version pypi Version conda Version

FAQ

Why does this exist?

The main purpose of this package is to make the default spaCy installation smaller and not force every user to download large data files for all languages by default. Lookups data is now either provided via the pre-trained models (which serialize out their vocabulary and lookup tables) or by explicitly installing this package or spacy[lookups].

When should I install this?

You should install this package if you want to use lemmatization for languages that don't yet have a pretrained model available for download and don't rely on third-party libraries for lemmatization โ€“ for example for Serbian or Turkish (see data files).

If you are training new models with spaCy, you should probably install this, since it contains lemmatization and normalization data for 25+ languages that is no longer included as part of the main spaCy library. In particular, you should install it if you're creating a blank model and you want it to include lemmatization and normalization data. Once you've saved out the model (e.g. via nlp.to_disk), it will include the lookup tables as part of its Vocab.

Is this package only for lemmatization?

This package used to only be for lemmatization, but it has been extended to include normalization data for many languages. As of v0.3.1 it also includes optional probability and Brown cluster data that used to be distributed with provided models in spaCy v2.2 but is no longer included in spaCy v2.3. In the future it may include other lookup lists and tables as well, e.g. large tokenizer exception files.

Running tests

This package now also includes all data-specific tests. The test suite depends on spaCy.

pip install -r requirements.txt
python -m pytest spacy_lookups_data

If you've installed the package in your spaCy environment, you can also run the tests like this:

python -m pytest --pyargs spacy_lookups_data

Bug reports and other issues

Please use spaCy's issue tracker to report a bug, or open a new thread on the discussion board for any other issue.