Word2vec models trained on English Wikipedia

This repository contains Word2Vec models trained on the full text of the English Wikipedia as downloaded in December 2021.

Preprocessing:

• lowercasing
• n-grams up to 4-grams were computed using Bouma 2009 (https://svn.spraakdata.gu.se/repos/gerlof/pub/www/Docs/npmi-pfd.pdf), min freq threshold of 10

Two models, trained with Gensim:

• wiki_300_5_word2vec --> dim 300, freq threshold 5
• wiki_300_50_word2vec  --> dim 300, freq threshold 50

Other hyperparameters set as follows: window=5, epochs=5, seed=1830, sg=1

Note:
Machine learning models trained on uncurated data inevitably learn hidden or obvious biases and as a result, the models shared with here might contain characteristics including sexism, racism, antisemitism, homophobia, and other such types of unacceptable biases. I encourage whoever is using these models to make sure such biases are actually removed before using them in production settings (see eg https://aclanthology.org/N19-1061/)