Transformers · Level 3 of 5
Value
An attention representation combined using attention weights.
It is distinct from the key used to calculate those weights.
Example
A weighted sum mixes source information vectors.
Listen to the definition and example
Audio transcript
Value. An attention representation combined using attention weights. It is distinct from the key used to calculate those weights. For example: A weighted sum mixes source information vectors.
Explore this concept
Why it matters
This helps you trace how token representations exchange and transform information.
Start with
Related concepts
Quick recall question
Try answering before looking back at the definition.