How an LLM Generates Text, One Token at a Time

A tiny, runnable look at how a pre-trained LLM turns a prompt into its next token.