Glossary · Phase 1 · Building with LLMs
Sampling
Picking the next token at random according to the model's probabilities, instead of always taking the top one.
Try it now
Watch the text grow by one word per pass. Each pass looks up the options for the last word, keeps the one with the highest probability and adds it to the end. It stops when "<end>" wins.
Learn it properly
What an LLM actually does · Next-token prediction, and why that produces useful behavior. About 25 minutes.
Other terms from the same lesson
- click-through rate: The share of people who saw a link or ad and clicked it: clicks divided by views.
- conversion rate: The share of people who went on to do the thing you wanted, such as buying, out of those who clicked.
- LLM: A large language model: a program trained on huge amounts of text to predict the next word, which lets it write, answer and reason.
- next-token prediction: The one task a language model is trained on: guess the next piece of text from everything before it.
- vocabulary: The full list of tokens a model knows.
Never trained an AI?
Teach a small one to write, right in your browser. It takes about 20 seconds and needs no account.