Explore Library
Slides / VideoIntermediate

Next-Token Prediction and Autoregressive Generation

LLMs generate text one token at a time, predicting the next token from everything seen so far.

Slide 1
1 / 6