Incorporate fasttext vectors in tf.keras embedding layer?

Question

Fasttext could handle OOV easily, i.e., it could be assumed that emb = fasttext_model(raw_input) always holds. However, I am not sure how I could build this layer i