Glossary

LLM (Large Language Model)

LLM (Large Language Model) is a neural network trained on trillions of tokens of text that generates human-like text in response to a prompt. ChatGPT, Claude, Gemini, Llama — all are LLMs. The foundation of the current AI boom.

How it works

An LLM predicts the next token (a word or part of a word) based on context. Despite the simple task, the scale and Transformer architecture allow it to "understand" complex instructions, write code, translate, and analyse.

Key concepts

Token — the smallest unit of text. GPT-4 processes ~4 characters = 1 token
Context window — the maximum amount of text the model "sees" at once
Temperature — a randomness parameter. 0 = deterministic, 1 = creative
Inference — generating a response (as opposed to training)

Limitations

Hallucinations (confident but false facts), knowledge capped at training date, limited context window, expensive to scale.