The MXnet installation pit in the English version of the hands-on deep learning book (1.6 version cpu installation)

MXNet installation pit

1. Description

Because I want to run the English version and learn the code of the deep learning recommendation system chapter, I have an environment.

Students who want to run the Chinese version and learn deep learning code do not need to read this blog .

2. First is the installation of the official website address

Click me to get Start

3. English version hands-on learning deep learning MXnet version 1.6 installation

First, the Chinese version of the library is used d2lzh, with the English version Shi d2l, not the same .

I installed the 1.6 version directly according to the English version MXnet（cpu）. The gpu version has been installed for a long time. I always report errors when guiding the library. I also hope that some big guys can comment and leave a message. Thank you very much . Click here for the website .

The installation of the cpu version is roughly as follows :

• First create a virtual environment in anaconda, py version 3.x.
• pip install mxnet==1.6.0
• pip install -U d2l ( -U: Upgrade all packages to the newest available version)
• Try the code in the recommended system chapter as unsuccessful (just follow the instructions on the book to guide a library)

The installation of the gpu version on the English official website is roughly as follows . For details, go to the English website. I tried it, and the 1.6 version of MXnet (gpu) directly reported an error with pip, it seems there is no corresponding gpu version.

I also tried to download the 1.2 version of mxnet that comes with anaconda, and install the d2l library. It is installed, but when I follow the guide mxnet and d2l in the English book, an error will be reported. It seems that the name np and npx cannot be . Checked it seems to be able to use np and npx from version 1.6 on windows.

English official website for installation:

# For Windows users
pip install mxnet-cu101==1.6.0b20190926

# For Linux and macOS users
pip install mxnet-cu101==1.6.0