# PyPI Mirror

# Introduction

PyPI (opens new window), a.k.a. the Python Package Index, is a centralized "repository of software for the Python programming language".

The PyPI Mirror is synchronized with pypi.org (opens new window) with bandersnatch (opens new window).


pip install --upgrade pip --index-url https://mirrors.sustech.edu.cn/pypi/web/simple
pip config set global.index-url https://mirrors.sustech.edu.cn/pypi/web/simple

That's all.

# Detailed Instructions

# 1. Confirm your python environment

The python and pip commands you're using may be

  • installed with package manager (e.g., apt-get, yum, etc.)
  • self-compiled
  • isolated by a python 3 venv (virtual environment)
  • equipped with Anaconda or installed in a conda virtual environment
  • bundled with Intel Parallel Studio
  • ...

You can execute the following commands to gain insight into the pip command that you are using.

which pip
pip --version

For Python 3 Users:

You may need to replace pip with pip3.

# 2. Configure Index-url

# For pip >= 10.0

pip config set global.index-url https://mirrors.sustech.edu.cn/pypi/web/simple
  • Use the https schema instead of http
  • Don't omit simple

# For older versions

You can upgrade to the latest version first:

# you may need root privilege
pip install --upgrade pip --index-url https://mirrors.sustech.edu.cn/pypi/web/simple

Or, you can stay with the old version and manually edit the configuration file with your text editor:

index-url = https://mirrors.sustech.edu.cn/pypi/web/simple

Path of the per-user configuration file:

  • Linux: $HOME/.config/pip/pip.conf
  • macOS: $HOME/Library/Application Support/pip/pip.conf or $HOME/.config/pip/pip.conf
  • Windows: %APPDATA%\pip\pip.ini

You may create a new file if it doesn't exist.

You can refer to pip user guide (opens new window) for per-virtualenv or site-wide configuration.

# Temporary Use

pip install package-name --index-url https://mirrors.sustech.edu.cn/pypi/web/simple

# Install Packages without Root Privilege

In an environment where you have no root access, like SUSTech Taiyi/Qiming HPC or your lab machines, you are still able to install the python dependencies you want.

# Option 1: Python 3 Venv (Suggested)

You can create multiple virtual environments in your home directory so that you can have dependencies of different versions for different projects.

# 1. Create a virtual environment
[user@host ~]$ python3 -m venv ~/venv-torch

# 2. Activate the virtual environment
[user@host ~]$ source ~/venv-torch/bin/activate

# 3. Now you are in the venv
(venv-torch) [user@host ~]$ pip install --upgrade pip  # optional
(venv-torch) [user@host ~]$ pip install torch torchvision
(venv-torch) [user@host ~]$ python my_awesome_network.py

# 4. Leave the virtual environment
(venv-torch) [user@host ~]$ deactivate
[user@host ~]$  # now you've come back

Essentially, this approach creates symbolic links for system-wide python.

Please refer to https://docs.python.org/3/library/venv.html (opens new window) for detailed usage.

# Option 2: Pip User Install

Suppose you're using Python 3.7 and you want to install Numpy. You can simply run:

pip install --user numpy  # you may need to use pip3

By default, numpy would be installed to ~/.local/lib/python3.7/site-packages/numpy-*. This method is not suggested if you are sharing an account with others.

Please refer to https://pip.pypa.io/en/stable/user_guide/#user-installs (opens new window) for detailed usage.

# Option 3: Conda virtual environment

If you need Python of different versions, for example, Python 3.4 for project A and Python 3.7 for project B, you can consider create several Conda environments.

Please refer to https://docs.conda.io/projects/conda/en/latest/user-guide/tasks/manage-environments.html (opens new window) for detailed usage.