← Tutorials2 / 22
What is an LLM?
What does 'predicting the next token' actually look like?Tap to reveal
"The cat sat on the" ─▶ model
│
mat ████████████ 0.41
floor ████ 0.12
sofa ███ 0.08
... (50k tokens)The model outputs a probability for every token in its vocabulary. Generation picks one of them (usually a likely one), appends it, and asks again.
The model never 'looks up' an answer: it continues text in the way that best matches what it learned.