Skip to content

Generative AI & LLMs

Next-token prediction

The training objective where a language model learns to predict the next token in a sequence.

Example

Given “The meeting starts at,” the model assigns probabilities to possible next tokens.

Why people use it

Teams use “Next-token prediction” when they need to choose and configure generative systems more effectively.

What you'll hear

“How does Next-token prediction affect the answer the model gives the user?”

Related terms