Learned Position Embeddings

What is Learned Position Embeddings? Why does it exist? Where does it appear in Positional Information?

Draw or describe a visual model of Learned Position Embeddings. What does it look like conceptually?

Work through a concrete manual example of Learned Position Embeddings with real numbers or pseudocode.

How is Learned Position Embeddings different from related concepts in Positional Information? What are the trade-offs?

Explain Learned Position Embeddings in plain language using an analogy. Could a non-expert follow it?