1.1 / THE MODEL

How LLMs work

A working mental model: token-by-token prediction, and why a model that keeps nothing between calls is the reason harnesses exist at all.

AUTHOR LEA DANG WORDS 0