Why does the TextVectorization layer return an empty tensor for the empty string instead of the index zero? Is there a way to return a tensor of zeros?
import