University of Michigan - Ann Arbor

An attention function is defined as a mapping from a query and a set of key-value pairs to an output, where the query, keys, values, and resulting output are all represented as vectors.

Vector Mapping Definition of Attention

In the formal definition of attention, how are the keys and values structured when provided as inputs to the mapping function?

True or False: In the formal definition of attention, the attention function maps from multiple queries simultaneously to produce a single output.

In the formal definition of an attention function, what mathematical representation is uniformly shared by the query, the keys, the values, and the resulting output?

Learn Before

Related