Overview
An Autoregressive large language model is designed to try to predict the next word (token) in a sequence, given the prior words in the sequence. Formally, this can be written as
{% \mathbb{P}(t | t_1,t_2,...,t_n) %}
(see conditional probability)