How a prompt travels through 28 layers
What happens inside an LLM between your prompt and its answer?
Between your prompt and its answer, an LLM runs your text through a stack of layers. Each layer adds to a working state. The answer comes from that state after the last layer.
