Comment by arnitdo

Isn't this the same thing as your token probability distribution? A set of likely tokens related to the input?