GLiNER stands for Generalist and Lightweight model for Named Entity Recognition. It fills a niche between traditional NER models which are limited to pre-defined classes of entities and full blown generative LLMs which are resource intensive and costly to run.
GliNER has spaCy integration which can be installed via pip install gliner-spacy which makes it very convenient to use in existing spacy pipelines. The repo also provides example code for fine-tuning the model.