Formula

Set of Superscript-Indexed Vectors ({qi[1],,qi[τ]}\{ \mathbf{q}_{i}^{[1]}, \dots, \mathbf{q}_{i}^{[\tau]} \})

The notation {qi[1],,qi[τ]}\left\{ \mathbf{q}_{i}^{[1]}, \dots, \mathbf{q}_{i}^{[\tau]} \right\} represents a set of vectors. Each vector, denoted by qi\mathbf{q}_i, is further identified by a superscript index enclosed in square brackets, such as [1][1] or [τ][\tau]. This indexing scheme is commonly used to represent a collection of vectors associated with a single element ii across multiple parallel contexts, for instance, the query vectors for the ii-th token across τ\tau attention heads.

0

1

Updated 2026-07-03

Contributors are:

Who are from:

Tags

Ch.2 Generative Models - Foundations of Large Language Models

Foundations of Large Language Models

Foundations of Large Language Models Course

Computing Sciences