Quantized word vectors that take 8x-16x less storage/memory than regular word vectors - View it on GitHub