Post Hoc Explanations of Language Models Can Improve Language Models

Open in new window