A Deep Dive into Large Language Models

Curious how ChatGPT and other LLMs actually work? This self-paced Socratic series takes you behind the scenes step by step—guided by an AI that helps you question, explore, and discover how these models learn. You’ll explore how models are trained, why they hallucinate, and how they learn to be useful. Created based on Andrej Karpathy’s Deep Dive into LLMs like ChatGPT.
Library
Course cover
Lesson 1
What Makes LLMs Different? image

What Makes LLMs Different?

You know what LLMs can do. But do you know what makes them so different from any AI that came before?
Start
Lesson 2
From Raw Internet to Quality Data image

From Raw Internet to Quality Data

Learn how the messy internet is transformed into clean data for training AI.
Start
Lesson 3
From Text to Tokens image

From Text to Tokens

Discover how AI transforms text into something it can learn from.
Start
Lesson 4
Neural Network I/O image

Neural Network I/O

See how a neural network learns to predict the next word.
Start
Lesson 5
Neural Network Internals image

Neural Network Internals

Peek under the hood and see how an AI's 'brain' really works.
Start
Lesson 6
From Prediction to Generation image

From Prediction to Generation

Learn how an AI goes from predicting one word to writing a whole story.
Start
Lesson 7
Watching AI Learn: From Gibberish to Text image

Watching AI Learn: From Gibberish to Text

Go behind the scenes and see what it takes to train a large language model from scratch.
Start
Lesson 8
Teaching AI to Chat image

Teaching AI to Chat

Discover how a text-predicting machine is taught to become a helpful conversational assistant.
Start
Lesson 9
Taming Hallucinations image

Taming Hallucinations

Ever seen an AI make stuff up? Learn why it happens and how we can teach it to be more truthful.
Start
Lesson 10
A Borrowed Identity image

A Borrowed Identity

Does an AI know who it is? Uncover the illusion of self and how LLMs get their identity.
Start
Lesson 11
Trial, Error, and Genius image

Trial, Error, and Genius

Discover how learning from mistakes can lead to mastery — even for machines.
Start
Lesson 12
Reinforcement Learning from Human Feedback image

Reinforcement Learning from Human Feedback

Discover how AIs learn what 'good' looks like when there’s no single right answer.
Start

See relevant courses