Showing cs.FLShow all
2 papers · 1 filter
cs.FL2026
Algebraic Decomposition Theory for Transformer Length Generalization
Andy Yang, Blerta Veseli, Corentin Barloy +5
Transformer-based language models are known to sometimes generalize to sequences longer than seen during training, but we lack a precise characterization of which tasks admit lengt…
cs.FL2024
Masked Hard-Attention Transformers Recognize Exactly the Star-Free Languages
Andy Yang, David Chiang, Dana Angluin
The expressive power of transformers over inputs of unbounded size can be studied through their ability to recognize classes of formal languages. In this paper, we establish exact…