LLM Fundamentals → Frontier

Understanding frontier LLMs can be easy.

From fundamentals to the cutting edge intuitively, with everything in one place.

Start learning View pricing

[ 01 ]173,702Words of curriculum

[ 02 ]48Interactive visualizations

[ 03 ]56Quiz questions, 9 quizzes

[ 04 ]29Sections across 9 chapters

Learners from

NortheasternNYUUniversity of WashingtonAI2

Learners

What existing learners say

[ 01 ]

Jeremy Wu

Master in Analytics, Northeastern University

i just learned one hot encoding in my python class this quarter, and you nailed it here:
I like how you show how AI is limited, mentioning the traditional one hot encoding and the solution such as Dense Embeddings for more complex problems.
Smooth explanations of embeddings and transformers, i think your embedding chapter is great!
I think those with a bit of understanding about AI will cruise through your course based on the first 3 chapters.
Simple explanations, not too wordy, colloquial so more entertaining for the readers, and not too long to lose their interest!

[ 02 ]

Ray Wen

Master in computer science, New York University

Intuitive AI Academy replaces the typical AI information overload with a streamlined, intentional curriculum.
It strips away the distractions, allowing you to focus on building a cohesive understanding of LLMs without losing the freedom to explore the details.

[ 03 ]

Jason Lee

masters student at University of Washington, student researcher at ai2

There's no doubt that AI and deep learning will have a tremendous impact on the world.
No matter you background, I believe it's never too late or too difficult for anyone to learn and fully understand the future we're building.
Intuitive Academy lowers the barriers to entry for AI systems, enabling everyone to truly comprehend AI from the core fundamentals to the latest state of the art breakthroughs.

§ 01 · Comprehension

Most AI explanations are either too vague or too hard.

You get the intuition first, then a visualisation you can run yourself, then the maths.

This is one of them. Try it.

A question goes in. Watch the model search everything it has been given and pick out the one word that answers it.

48Interactive explainers, built by hand

Fig. 2 — Query–Key Retrieval, from the Attention chapterOpen full size →

The maths is still in there

We do not skip the equations. You just get to see what they do before you have to read them, so they make sense when you get there.

Built in order

Tokenization, then embeddings, then attention, then what you get from stacking them. Every chapter only uses what an earlier one taught you.

All of it in one place

173,702 words and 410 diagrams, updated constantly. Everything is here, so you are not hunting across the web for the next piece.

§ 02 · Progress

Most people quit around chapter two. This is built so you don't.

Finish sections, pass quizzes, earn XP for both, and see where you land on a leaderboard that starts level every Monday.

29Sections to finish

56Questions across 9 quizzes

XP for all of it

10 for a section, 15 for every question you get right, 100 for finishing a chapter.

A week on the site

MONSection · Tokenization+10

MONSection · The Embedding Layer+10

TUEQuiz · Architecture · 7 of 8+105

WEDSection · Positional Encoding+10

THUChapter finished · Architecture+100

FRIWeek active+25

Week total+260

This weekResets Monday

01Mei L.640

02Daniel K.585

03You260

04Priya R.295

Everyone starts level

Everyone starts over on Monday

The board wipes every week, so it makes no difference whether you joined last year or last night. This one is anyone's. And if you would rather not be on it at all, opt out and nobody sees you.

A streak that survives a bad week

Miss a day and nothing happens. Miss three and it is still fine. Come back before the week is out and your streak is right where you left it.

§ 03 · Who it is for

Three kinds of people get the most out of this.

Engineers

Shipping features on top of LLMs, and tired of guessing what the model is doing underneath.

Students and researchers

Getting current quickly, without working through forty papers to find the six that matter.

Founders and operators

Deciding what to build, and needing a straight answer on what these models can and cannot do.

§ 04 · Curriculum

Course contents

Two courses, 29 sections. Each one only uses what an earlier one taught you.

LLM Fundamentals

3 chapters

Architecture9

01Introduction

02Tokenization

03The Embedding Layer

04Positional Encoding

05Attention

06Layers of Understanding

07Learning to Predict

08Instruction Tuning and RLHF

09GPT-2 from Scratch

Pre-Training8

01Overview

02Training Objectives and Architectural Details

03Scaling Laws and Optimization

04Training Data Engineering

05Training Infrastructure and Systems

06Advanced Pretraining Objectives

07Evaluation During Pretraining

08Case Study - LLaMA 3

Post-Training5

01Overview

02Supervised Fine-Tuning

03Preference Optimization

04Tools and Safety Tuning

05Case Study on Tulu 3

LLM Advanced

6 chapters

Distillation1

01Distillation

Linear Attention1

01Linear Attention

LoRA1

01LoRA

Mixture of Experts1

01Mixture of Experts (MoE)

Optimizers1

01Optimizers

Reinforcement Learning2

01RL Fundamentals

02RLHF

§ 05 · The reader

Inside the reader

This is the LoRA chapter, from the advanced course.

A course chapter on LoRA, showing the sidebar, the chapter contents rail, and hand-drawn diagrams of low-rank adaptation — Fig. 1 — LoRA: Low Rank Adaptation

§ 06 · Pricing

Accessible pricing for everyone

No tiers and no per-course pricing. Cancel whenever.

Monthly

$12/month

Billed monthly

All 173,702 words across 29 sections
48 interactive visualizations
56 quiz questions across 9 quizzes
Progress tracking, bookmarks and search
New content added constantly
Learning community access

YearlyBest value

$120/year

Save $24 · $10/month

All 173,702 words across 29 sections
48 interactive visualizations
56 quiz questions across 9 quizzes
Progress tracking, bookmarks and search
New content added constantly
Learning community access

Don't have an account? Start learning to get started.