#formallanguage

LLMs hit formal reasoning ceiling; Chomsky Hierarchy reveals efficiency gap

New benchmark shows large language models struggle with structured complexity tasks and require prohibitive compute to achieve reliability in formal reasoning.