Biasless Language Models Learn Unnaturally: How LLMs Fail to Distinguish the Possible from the Impossible.

Imry Ziv, Nur Geffen Lan, Emmanuel Chemla

Browse the full EACL paper archive.