Go to top

Introducing spaCy v3.4

We’re pleased to publish v3.4 of the spaCy Natural Language Processing library. spaCy v3.4 brings typing and speed improvements along with new vectors for English pipelines and new trained pipelines for Croatian. This release also includes prebuilt linux aarch64 wheels for all spaCy dependencies distributed by Explosion.

Typing improvements

spaCy v3.4 supports pydantic v1.9 and mypy 0.950+ through extensive updates to types in Thinc v8.1.

Speed improvements

  • For the parser, use C saxpy/sgemm provided by the Ops implementation in order to use Accelerate through thinc-apple-ops.
  • Improved speed of vector lookups.
  • Improved speed for Example.get_aligned_parse and Example.get_aligned.

Trained pipelines

New trained pipelines

v3.4 introduces new CPU/CNN pipelines for Croatian, which use the trainable lemmatizer and floret vectors. Due to the use of Bloom embeddings and subwords, the pipelines have compact vectors with no out-of-vocabulary words.

PackageUPOSParser LASNER F
hr_core_news_sm96.677.576.1
hr_core_news_md97.380.181.8
hr_core_news_lg97.580.483.0

Pipeline updates

All CNN pipelines have been extended with whitespace augmentation.

The English CNN pipelines have new word vectors, which improve the NER performance and update the vectors with words like “AirTags”, “Brexit”, “covid” and “doomscrolling”:

PackageModel VersionTAGParser LASNER F
en_core_web_mdv3.3.097.390.184.6
en_core_web_mdv3.4.097.290.385.5
en_core_web_lgv3.3.097.490.185.3
en_core_web_lgv3.4.097.390.285.6

New in the spaCy universe

Many cool new plugins, extensions, pipelines and tutorials have been added to the spaCy universe since v3.3:

Aim-spacyAn Aim-based spaCy experiment tracker.
AsentFast, flexible and transparent sentiment analysis.
spaCy fishingNamed entity disambiguation and linking on Wikidata in spaCy with Entity-Fishing.
spacy-reportGenerates interactive reports for spaCy models.
View the spaCy universe

Resources