3Blue1Brown’s “Neural networks” playlist

About the resource

3Blue1Brown, presented by Grant Sanderson, is one of the best-known educational maths channels, famous for precise animations that make difficult ideas visible and understandable. Its “Neural networks” playlist is one of the best things to watch for anyone who wants to understand how neural networks really work, rather than just memorise their vocabulary. The playlist starts from a simple question – what is a neural network? – then explains how a network learns from examples, and in its newer videos reaches large language models, the transformer architecture and the attention mechanism behind tools such as ChatGPT.

The series’ strength is that it shows you what happens inside the network instead of stopping at the equations: you watch the weights change, the error shrink, and words turn into vectors of numbers. That visual understanding stays with you for a long time and makes advanced books and explanations much easier to read later.

What you will find in the playlist

• Chapter one: what a neural network is, through the example of recognising handwritten digits.
• Gradient descent: how the network gradually adjusts its weights to reduce error.
• Backpropagation: an intuitive explanation first, then a video on the calculus for those who want to go deeper.
• A short video explaining large language models briefly for non-specialists.
• Chapters on transformers and the attention mechanism, and how models store facts.

Who it suits

It suits university students and graduates interested in artificial intelligence, and secondary-school students who are strong in maths. The first videos can be followed without much mathematical background, while the calculus videos need some knowledge of derivatives and matrices – you can skip them on a first viewing. The videos are in English, and many have subtitles in various languages.

How to use it, step by step

1. Watch the videos in order, and pause when a new idea appears to try to predict the next step.
2. After each video, write a short paragraph on what you understood, in your own words.
3. If the maths in the calculus videos is hard, watch them again after revising derivatives.
4. Connect what you watch to practice: build a small network with a library such as PyTorch after chapter four.
5. Watch the transformer videos after reading “What are large language models?” on the academy’s blog.

Before you start

The playlist’s videos appear on this page, and you can watch any of them directly in YouTube’s privacy-enhanced player. In the artificial intelligence course at Badr Academy Education we build on these ideas with practical applications. The content and its rights belong to 3Blue1Brown.

Links

• The resource in the library
• The original source – https://www.youtube.com/playlist?list=PLZHQObOWTQDNU6R1_67000Dx_ZCJB-3pi
• But what is a neural network? | Deep learning chapter 1
• Gradient descent, how neural networks learn | Deep Learning Chapter 2
• Backpropagation, intuitively | Deep Learning Chapter 3
• Backpropagation calculus | Deep Learning Chapter 4
• Large Language Models explained briefly
• Transformers, the tech behind LLMs | Deep Learning Chapter 5
• Attention in transformers, step-by-step | Deep Learning Chapter 6
• How might LLMs store facts | Deep Learning Chapter 7
• But what is cross-entropy? | Compression is Intelligence Part 2
• But how do AI images and videos actually work? | Guest video by Welch Labs

The videos are watched in YouTube’s player on the resource page or on YouTube; YouTube’s terms do not allow downloading them, so this sheet carries the resource’s description and links.