Square chestnuts from au Fei Temple


No. | public QbitAI qubit reports

Teacher Li Mu’s new book is online (again).

It is the Chinese preview version of Hands-on Deep Learning.

Compared to version 0.7, which was released in July, almost every chapter has visible changes.

For example, in the chapter of computer vision, a new section of multi-scale object detection has been added.


What doesn’t change is that as an interactive textbook, in addition to the book itself, it also provides easy code implementation and a forum full of experts to exchange skills.

Most importantly, it’s a book from scratch. It can be eaten with or without a base.

There are many ways to open a book

As usual, take a look at the author list:


In addition to Li Mu, the chief scientist, the other three are also from Amazon.

(These four were joined by more than 100 contributors.)


Now look at the content. According to the book, the textbook can be divided into three parts:

Part I (Chapters 1-3), preparatory work and basics.


Part 2 (Chapters 4-6), Modern deep learning techniques.


Part 3 (Chapters 7-10), Computing performance and applications.


Let’s use a diagram to show the context of the book, which looks like this. The direction of the arrow, which represents the previous chapter, is helpful for the understanding of the next chapter.


In addition, this book in addition to text version, and Jupyter notepad, each section can be run.

The code is also free to change.

As a result, the formulas in the book have not only words and diagrams, but also code that can be demonstrated intuitively.

It looks like an all-round friendly experience.

Beautiful words:


A friendly illustration:


Gentle code:


In addition, when you don’t want to study, you can go to the community (the portal is at the bottom of the text) to observe the works of the Great God, for motivation.

Alternatively, sharing your own problems and watching others’ can give you some confidence.

Love of learning each teenager, whether drool?

There are also classes with the same name

In addition to textbooks, Hands-on deep Learning is also the name of a set of courses.

Classes started in September last year, and the first season officially ended in February this year.

There are 19 classes, produced by the MXNet team. Li Mu is the author of MXNet.

The course content is based on the textbook and uses Gluon, the latest front-end for Apache MXNet 159, as a development tool.


All the videos are out now.

While attending class, while reading, while there is homework to write, Li Mu teacher will personally encourage everyone. You must be very happy.

It was posted on October 26 last year

Aren’t you going to start?


As for the differences between the new textbook and the 0.7 version, here is not much to mention.

But it’s 16 pages thicker than it used to be. There’s more to learn. Aren’t you happy?

New textbook Portal:

Hands-on Deep Learning – Hands-on deep learning document​zh.diveintodeeplearning.org

(this is old: http://zh.gluon.ai/toc.html)

Jupyter Notebook Portal:

http://zh.diveintodeeplearning.org/d2l-zh.zip​zh.diveintodeeplearning.org

Portal of same name course:

https://discuss.gluon.ai/t/topic/753​discuss.gluon.ai

Forum portal:

https://discuss.gluon.ai/​discuss.gluon.ai

– the –

Welcome to our column: Qubit – Zhihu column

Sincere recruitment

Qubit is looking for editors/reporters to work in Zhongguancun, Beijing. Looking forward to the talented and enthusiastic students to join us!

For more details, please reply “Wanted” on the QbitAI chat screen.

Qubit QbitAI· Head number signing author

վ’ᴗ’ ի Tracks new developments in AI technology and products