Datasetpublic

almogtavor/google-analogy-dataset

Google Analogy Dataset (Columnized) This is a columnized version of the Google analogy dataset by Mikolov et al. (2013): https://github.com/nicholas-leonard/word2vec/blob/master/questions-words.txt The dataset contains word analogy questions grouped by subjects such as: capital-common-countries (e.g., Athens Greece Tokyo Japan) currency (e.g., USA dollar Japan yen) gram3-comparative (e.g., big bigger cold colder) The original dataset is widely used in word embedding evaluation… See the full description on the dataset page: https://huggingface.co/datasets/almogtavor/google-analogy-dataset.

sourceHugging Faceupdated 1y agoView on Hugging Face

2likes83downloads

Dataset Card

Google Analogy Dataset (Columnized)

This is a columnized version of the Google analogy dataset by Mikolov et al. (2013): https://github.com/nicholas-leonard/word2vec/blob/master/questions-words.txt

The dataset contains word analogy questions grouped by subjects such as:

—capital-common-countries (e.g., Athens Greece Tokyo Japan)
—currency (e.g., USA dollar Japan yen)
—gram3-comparative (e.g., big bigger cold colder)

The original dataset is widely used in word embedding evaluation, including in fastText.