Attention as "Look Back When Needed"

What is Attention as "Look Back When Needed"? Why does it exist? Where does it appear in Attention?

Draw or describe a visual model of Attention as "Look Back When Needed". What does it look like conceptually?

Work through a concrete manual example of Attention as "Look Back When Needed" with real numbers or pseudocode.

Implement Attention as "Look Back When Needed" from scratch in Python. Keep it minimal — just enough to see it work.

Try Attention as "Look Back When Needed" on a real example. Vary the inputs and observe what changes.

How is Attention as "Look Back When Needed" different from related concepts in Attention? What are the trade-offs?

Explain Attention as "Look Back When Needed" in plain language using an analogy. Could a non-expert follow it?

Build a small project or demo that uses Attention as "Look Back When Needed" as a core component.