semantics

Embeddings: how AI turns words into directions in space Lesson

Before a model can reason about a word, an image, or a paragraph, it has to turn it into a list of numbers -- a vector. Embeddings are how meaning gets encoded as position in space, so that closeness becomes similarity.