LLM Basics

Start here if you're new to large language models. Each page is short and focused on the minimum mental model you need to use LLMs well in code.

Start here

• What is an LLM — the next-token-prediction loop that underlies everything else.
• Tokens — what the model actually sees, and why it matters for prompting and pricing.

Going further

• Sampling — turning the next-token distribution into one token (temperature, top_p, top_k).
• Context Window — the token budget that caps every call.

How the pages relate

Read What is an LLM first — it introduces the loop. The other three zoom into one specific piece of that loop:

flowchart LR
    W[What is an LLM<br/>the next-token loop]
    W --> T[Tokens<br/>what goes in and out]
    W --> S[Sampling<br/>how each token is picked]
    W --> C[Context Window<br/>how much fits per call]