koaning/whatlies

Toolkit to help understand "what lies" in word embeddings. Also benchmarking!