On the Anatomy of Attention
This provides a foundational framework for understanding and comparing attention mechanisms in machine learning, though it is incremental in formalizing existing knowledge.
The paper introduces a category-theoretic diagrammatic formalism to systematically relate and reason about machine learning models, focusing on attention mechanisms to translate folklore into mathematical derivations and construct a taxonomy, and as an empirical example, identifies anatomical components of attention to explore variations.
We introduce a category-theoretic diagrammatic formalism in order to systematically relate and reason about machine learning models. Our diagrams present architectures intuitively but without loss of essential detail, where natural relationships between models are captured by graphical transformations, and important differences and similarities can be identified at a glance. In this paper, we focus on attention mechanisms: translating folklore into mathematical derivations, and constructing a taxonomy of attention variants in the literature. As a first example of an empirical investigation underpinned by our formalism, we identify recurring anatomical components of attention, which we exhaustively recombine to explore a space of variations on the attention mechanism.