Positional Encoding
Domain: NLP Fundamentals
🎯 Motivation
Adds position information to tokens since attention has no built-in sense of order.
📋 Concrete Example
The model adds a position pattern to each token: token 1 gets pattern A, token 2 gets pattern B, etc.
🔗 Analogy
Like numbering the pages of a book - the text on each page needs a page number so you know the order.