Computational linguistics

An autoregressive language model assigns a probability to the token sequence x₁, x₂, x₃. Which factorization matches this approach?