What Formal Languages Can Transformers Express? A Survey
It clarifies the fundamental capabilities and limits of transformers for researchers in NLP and theoretical ML, but is incremental as it synthesizes existing work.
The paper surveys theoretical research on the formal language capabilities of transformers, documenting diverse assumptions and providing a unified framework to harmonize contradictory findings.
As transformers have gained prominence in natural language processing, some researchers have investigated theoretically what problems they can and cannot solve, by treating problems as formal languages. Exploring such questions can help clarify the power of transformers relative to other models of computation, their fundamental capabilities and limits, and the impact of architectural choices. Work in this subarea has made considerable progress in recent years. Here, we undertake a comprehensive survey of this work, documenting the diverse assumptions that underlie different results and providing a unified framework for harmonizing seemingly contradictory findings.