No. 240 of 500 No. 24 in Deep Learning

The StatQuest Illustrated Guide to Neural Networks and AI

With Hands-On Examples in PyTorch!!!

By Josh Starmer

Published
January 6, 2025
Print length
363 pages
Reading level
Intermediate
ISBN-13
979-8303440616
Publisher
Independently published
Written for
Students, analysts and career-switchers who want to understand how neural networks and transformers actually work.

Synopsis

Josh Starmer, the former genetics researcher behind the StatQuest YouTube channel, follows his StatQuest Illustrated Guide to Machine Learning with a self-published guide of roughly 360 pages to the neural networks behind modern AI. Every concept is drawn in his signature cartoon style, with numbers flowing through small example networks so readers can watch weights, biases and activation functions at work. The book starts with a simple network fitting a shape to data, then builds through backpropagation, multiple inputs and outputs, ArgMax and SoftMax, and cross entropy before reaching convolutional networks for image classification.

Later chapters turn to sequences: recurrent networks and long short-term memory units applied to stock prediction, then word embeddings, seq2seq encoder-decoder models for translation, attention and transformers. The closing chapters distinguish decoder-only transformers, the design behind ChatGPT-style text generation, from encoder-only transformers used for classification and clustering, and a section on matrix algebra connects the drawings to the matrix operations real code performs. PyTorch exercises accompany the main chapters, showing how to implement each network step by step. A reviewer at NSF Unidata praised the book's balance of accessibility and technical depth. Readers finish with an intuitive, mechanical understanding of how large language models work and the coding basics to experiment further.

Scorecard

Editorial score

How it scored

Editorial score: 74.0 / 1005 weighted criteria

  • Influence30%

    19.5/30
  • Authority25%

    17.5/25
  • Practical value20%

    16.0/20
  • Currency15%

    12.0/15
  • Readability10%

    9.0/10
  • Editorial score

    74.0/100

How the rank was computed

Rank No. 240 of 500 comes from a composite of 55.5: observable signals 22.1 of 55 points plus editorial review 33.4 of 45 points.

All 11 inputs
InputPoints
Observable signals22.1/55
Reach (Amazon ratings volume)8.1/16.5
Consensus across best-of lists2.8/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted6.9/8.3
Amazon sales rank1.5/5.5
Recency2.8/2.8
Editorial review33.4/45
Author authority7.9/11.3
Influence8.8/13.5
Practical value7.2/9.0
Readability4.1/4.5
Currency5.4/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

Starmer's illustrated, step-by-step approach makes backpropagation, attention and transformers understandable, with PyTorch code to cement each idea.

Strengths

  • Step-by-step illustrations that build real intuition
  • Covers CNNs, LSTMs, attention and transformers
  • Hands-on PyTorch exercises for each concept

Keep in mind

The playful style and repetition can feel slow to experienced engineers, and the book covers neural networks only, not broader machine learning or deployment.

Key takeaways

  1. Backpropagation is the chain rule applied repeatedly, nudging every weight and bias in the direction that most reduces the network's error.
  2. Word embeddings turn words into numbers so that words used in similar contexts end up with similar values a network can process.
  3. Attention lets a model weigh every input word when producing each output, solving the memory bottleneck of earlier encoder-decoder translation models.
  4. Decoder-only transformers generate text one token at a time, while encoder-only transformers produce representations suited to classification and clustering.

Notable facts

  • Thomas Martin, an AI/ML software engineer at the NSF Unidata Program Center, reviewed the book and wrote that it strikes an excellent balance between accessibility and technical depth. Source (opens in a new tab)

About the author

Josh Starmer

Josh Starmer is the founder and CEO of StatQuest, whose widely followed YouTube channel teaches statistics and machine learning. He earned a PhD at North Carolina State University and spent more than a decade at the University of North Carolina at Chapel Hill, first as a postdoctoral researcher and then as an assistant professor, applying statistics to genetic data. He wrote The StatQuest Illustrated Guide to Machine Learning.

Questions

What does The StatQuest Illustrated Guide to Neural Networks and AI cover?
It moves from the basics of how a neural network fits data through backpropagation, SoftMax, cross entropy and convolutional networks, then to recurrent networks, LSTMs, word embeddings, seq2seq translation, attention and transformers. The final chapters explain decoder-only transformers used for text generation and encoder-only transformers used for classification, with PyTorch exercises throughout.
Do I need math or coding experience for the StatQuest neural networks book?
Very little math is assumed, because Starmer works through calculations with pictures and small numbers, and readers can understand the concepts without running any code. To complete the PyTorch tutorials, basic Python helps. Some readers report that the material demands steady attention despite the friendly tone, so it rewards working through chapters in order.
Is The StatQuest Illustrated Guide to Neural Networks and AI worth reading in 2026?
Yes. Published in 2025, it explains the transformer architecture that still underpins today's large language models, and its focus on intuition means the material ages slowly. It will not teach fine-tuning, retrieval or deployment of modern LLMs, so practitioners should treat it as a foundation and move on to engineering-focused books afterward.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.