Neural Networks and Transformers: Free Course

About this course

Build a visual understanding of neural networks and transformers in this free 3Blue1Brown course. The classroom preserves the seven numbered deep-learning chapters, the short LLM introduction, the cross-entropy recording and the guest explanation of AI images and videos. All ten source videos remain available in the published sequence.

Course at a glance

What you’ll learn

• Explain the purpose of weights, activations and layers in the introductory network.
• Connect gradient descent and backpropagation with a model’s learning process.
• Explore embeddings, transformer structure and attention through visual explanations.
• Recognize how the later recordings extend the core sequence into cross-entropy and generative imagery.

Skills you’ll gain

• Neural-network concepts
• Gradient descent
• Backpropagation
• Transformers
• Attention
• Visual mathematical reasoning

Course Overview

The first four numbered chapters establish a connected foundation: what a neural network represents, how learning can be viewed through gradient descent, and how backpropagation distributes the effect of a change. The short LLM introduction then provides context for Chapters 5–7, which turn to transformers, attention and the possible role of internal representations. The additional cross-entropy video and guest explanation of AI imagery remain part of the full source playlist. The course is primarily conceptual and mathematical rather than a software-development bootcamp. It gives you a stronger way to interpret the ideas behind models, but does not promise that watching the recordings will train a production model or explain every modern architecture.

How to study the course

• Study Chapters 1–4 in order. Sketch a small network and explain the role of a parameter before trying to follow every symbolic detail.
• Watch the short LLM introduction and Chapters 5–7. Keep a separate note for embeddings, attention and feed-forward processing so their purposes do not blur together.
• Complete the cross-entropy and guest-video additions. Relate each new idea to the core sequence and record where the explanation is conceptual rather than a step-by-step coding task.

Use the searchable classroom lesson list to revisit a specific explanation. Pause after a useful demonstration, reproduce a small example and explain the result in your own words. Keep a short note of what worked, what changed and what you still need to understand before moving on.

Who is this for?

This course suits mathematically curious learners, students and developers who want a clearer understanding of deep-learning concepts. It is especially useful when model terminology is familiar but the connection between calculations and behavior remains difficult to visualize.

Before you start

Comfort with algebra, functions and vectors is helpful; derivative knowledge supports the backpropagation mathematics. No paid AI service is required to view the lessons. Coding projects, model-training resources and assessed certification are not included in this classroom.

Put the lessons into practice

Draw a small layered network with a few inputs, hidden units and outputs. Explain in your own words how a forward calculation differs from the adjustment of its parameters. Then make a separate concept map for a transformer, showing where embeddings, attention and feed-forward processing belong. Choose one visual explanation from the course and recreate a simplified sketch without the video playing. Mark any point you cannot explain confidently and return to its source chapter. Finish with a short paragraph describing what the course helped clarify and what would require practical coding study next.

This optional activity is provided by WikiFree for independent practice. It is not a graded assignment from the instructor. Keep your working notes, check the result against the task you defined and return to the relevant lesson when an assumption or technique is unclear.

Certificate / Assessment

Watching the public videos and using WikiFree progress tracking does not award a certificate, exam voucher, academic credit or instructor assessment.

Use your practice output to review what you can now explain and repeat independently. Completion tracking records your study progress; it does not replace an assessed qualification or professional experience.

Frequently asked questions

Provider: 3Blue1Brown. Videos play from the 3Blue1Brown channel on YouTube.