Definition

Decoder-Only Transformer

A decoder-only Transformer uses just the generating half of the original architecture, reading text left to right and predicting the next token. BERT took the encoder and became a reader; GPT took the decoder and became a writer. The training objective needs no labels, only raw text, so the whole internet becomes training data. This choice is why modern LLMs are general-purpose and scale so well.

Decoder-Only Transformer diagram from Intelligence at Scale
Diagram from chapter 3, Attention Is All You Need

Explained in

Chapter 3: Attention Is All You Need Free

The paper behind every major AI model today. Google invented it. Then watched someone else ship it.

Related terms

This is one term. The chapter is the argument.

Intelligence at Scale: 22 chapters, 65,000 words, 80-plus diagrams. Kindle, paperback and hardcover on Amazon.

Buy on Amazon.com
← All terms