Training Data
Historical Irish FastText models were trained on Old, Middle, Early Modern, Classical Modern and pre-reform Modern Irish texts from St. Gall Glosses, Würzburg Glosses, CELT and the book subcorpus Historical Irish Corpus. The training data spans ca. 550 — 1926 and covers a wide variety of genres, such as bardic poetry, native Irish stories, translations and adaptations of continental epic and romance, annals, genealogies, grammatical and medical tracts, diaries, and religious writing. Due to code-switching in some texts, the models have some Latin in the vocabulary.
Available Models
There are 3 models in this familily:
- Cased, 119 630 words:
historical_irish_cased_ft_100_5_2.txt
- Lowercase, 112 230 words:
historical_irish_lower_ft_100_5_2.txt
- Lowercase with initial mutations removed, 99 485 words:
historical_irish_lower_demutated_ft_100_5_2.txt
All models are trained with the same hyperparameters (emb_size=100, window=5, min_count=2, n_epochs=100
) and saved as KeyedVectors
(see Gensim Documentation).
Usage
from gensim.models import KeyedVectors
from huggingface_hub import hf_hub_download
model_path = hf_hub_download(repo_id="ancatmara/historical-irish-ft-vectors", filename="historical_irish_lower_demutated_ft_100_5_2.txt")
model = KeyedVectors.load_word2vec_format(model_path, binary=False)
model.similar_by_word('coíca')
Out:
>>> [('coícat', 0.6620370149612427),
('coícait', 0.6584151983261108),
('coíctu', 0.550497829914093),
('trícha', 0.537602424621582),
('cóeca', 0.531631350517273),
('cóecta', 0.5148215889930725),
('cóecait', 0.5108019113540649),
('tríchad', 0.5059043765068054),
('tríchaid', 0.5049244165420532),
('cóecat', 0.5042815804481506)]