en_core_web_trf is spaCy’s English transformer pipeline: a large RoBERTa-based model used here to support high-accuracy coreference resolution with Coreferee. The trf suffix marks transformer integration; its size and pinned dependencies make an isolated environment practical. section 11.5.1
Load the pipeline for coreference resolution
Install the English model and Coreferee, then load the transformer pipeline and attach the resolver: `nlptrf = spacy.load('en_core_web_trf')`; `nlptrf.add_pipe('coreferee')`. Processing text exposes `doc._.coref_chains`, which can link mentions such as “Gebru” with repeated “she” references and “advice” with “it”. section 11.5.1
$ python -m spacy download en_core_web_lg #1 $ python -m spacy download en_core_web_trf #2 $ python -m coreferee install en
The installation commands show the required English transformer model, larger language model, and Coreferee language resources. section 11.5.1
