Large Language Model (LLM)

Definition

A large language model (LLM) is a neural network trained to predict the next token in an input sequence.

For example, given “All that glitters …”, an LLM can predict the continuation ”… is not gold.”

At a high level, this next-token prediction capability is the foundation of modern language models. LLMs learn patterns from large amounts of text and use those patterns to generate responses.

Why It Matters

Understanding LLMs is the starting point for understanding modern generative AI applications. For a PM, the LLM is the core model capability everything else in an AI product gets built around.