These weights are extracted and they act as word embeddings. This is part 2/2 for training word embeddings. In part 1 we understood the theory behind training word embeddings.