spacy-models

Why we included this project

Building production NLP pipelines in Python usually means needing the trained pipelines that spaCy runs on, and this is where Explosion publishes them. The models are too big to sit in the repo as binaries, so they ship as pip-installable wheels and source archives on GitHub Releases; you grab one with `python -m spacy download en_core_web_md` or pin it in requirements.txt. The naming convention carries real information: `en_core_web_md` tells you the language, which components are included (tagger, parser, lemmatizer, NER), the kind of text it was trained on, and whether word vectors come along. So the small, medium, and large builds are mostly a choice about whether you need those vector tables. Model versions track the spaCy release they target, which keeps compatibility checks straightforward when you upgrade the library.