← All coursesTRACK A · AI FROM ZERO · A1

The Transformer idea, in plain words

A complete beginner's guide to how many modern language models use context. Six short lessons, invented shop examples, and quizzes that explain each answer. No math or code needed.

$79 one-time

Lesson 1 is free to read. The full six-lesson course is available for $79. After checkout, use the course-start link provided by Stripe to work through the lessons. These are unlisted links, not account-based access control.

Secure checkout by Stripe. One-time purchase; tax may be added based on your address. Lessons are unlisted but not locked.
THE COURSE

Six lessons, one clear idea

About 45 minutes total
  1. 01

    Why one word can mean two things

    7 min read · Free sample

    Find the clues that change a word's meaning before you trust an answer.

    Read free lesson ↗
  2. 02

    Paying attention, the dinner-table version

    7 min read · Unlisted course lesson

    See how words connect to useful clues, and why attention is not a truth detector.

  3. 03

    Reading the whole page at once

    7 min read · Unlisted course lesson

    Connect distant clues in an input without confusing reading with output generation.

  4. 04

    How ChatGPT writes one word at a time

    7 min read · Unlisted course lesson

    Understand tokens, fluent drafts, and the need to check source facts.

  5. 05

    The 2017 paper that started it all

    7 min read · Unlisted course lesson

    Meet the original Transformer research and the translation problem it tackled.

  6. 06

    Why this unlocked everything after

    8 min read · Unlisted course lesson

    Separate model, instructions, tools, and agents; put checks around action.

What this course is and is not

These are self-guided lessons with practice quizzes. There is no live teacher, certificate, or account-based access control yet. The shop examples are invented to make the ideas easier to see.