Lemmy3 is an experimental fork of Lemmy
It has been refactored in object-oriented manner, the codebase is extended with type-hints and spacy-compatible serialization, and a simple frequency-based disambiguation method is added. The tool comes with a command line interface (lemmy
) for training lemmatizer models.
To get started using the lemmatizer, install it from PyPI:
pip install lemmy3
To train a model one can use the packaged CLI. (Note that your data must be in CONLLU format)
lemmy train train.conllu model.bin
Evaluation of the model is simply possible by issuing the following command.
lemmy evaluate model.bin test.conllu
import lemmy
nlp.add_pipe("lemmy3", before="parser")