v1.0.0

• Supports (and requires) spaCy v3.0 ♻️ ⬆️
• Enables morphological analysis (token.morph) ✨
• Full serialization support 🚸

v0.3.2

Sets spaCy version range to spacy>=2.1.0,<3.0.0 to ensure proper install when spaCy v3 releases

v0.3.1

• Add multiprocessing hotfix (#21)
• Extract __version__ (enables #22)

v0.3.0

• Adds support for pretokenized and presegmented text (#19)
• Enables Language.pipe processing (#20)

v0.2.1

• Adds three new models trained on UD 2.5 (#16)
• Adds proper support for multiword tokens (#17) and fixes related issues (#14 and #15)
  Thanks to @KoichiYasuoka for the contribution 😄

v0.2.0

• Updates default language models (trained on UD 2.5)
• Fixes #10
• Serialization works properly

v0.1.0

• New default language models (fixes #7)
• Adds GitHub Actions (#8)

v0.0.4

Fixes issue #4: Use MISC field from CoNLL-U Format to more accurately tokenize text

v0.0.3

Fixes issue #3: Use only ISO 639-3 language codes
Thanks to @KoichiYasuoka for the pull request!

v0.0.2

Fix issue #1: Add UDPipe model loading from disk