Skip to content
AImpact
IT EN
Models Intermediate Also known as: Autoregressivo

Autoregressive

A model that generates a sequence one element at a time, each time using the previous output as part of the new input.

ShareLinkedInX

In practice

It is how every GPT-style LLM works: each new token depends on all the previous ones. It explains why generation is inherently sequential and hard to parallelize, and is the reason behind tricks like speculative decoding to speed it up.

Related terms

Causal Mask Decoder-only Transformer LLM Greedy Decoding

Seen in the wild

3 entries mentioning it
  1. EMU3: a single transformer for text, images, and video
    High
  2. SoundStorm: Google generates 30 seconds of natural dialogue in half a second
    High
  3. Speculative Decoding: 2-3x LLM inference speedup without changing output
    High
← All terms