Enhancing the Rationale-Input Alignment for Self-explaining Rationalization

Open in new window