1, the virtual environment is doing with?
I upload a cuda computer, graphics drivers, cudnn supporting documents such as a pile, and then rely on the gpu cuda driver installation and test version tensorflow2.0, I wonder why, every time I finish tf2 computer program will be stuck with and I could change the graphics card hardware related, but I do not want to unload heavy equipment, in case of a bad computer environment collapse I have to install it again from scratch, and consequently there is a computer system, think about it hard, so I plan to install a cpu debug version of tf2.0 first look. So this time the virtual environment is a good choice.
2, how to install?
I environment:
python --version #python3.5.2
pip --version #pip 19.3.1
sudo pip install virtualenv sudo pip install virtualenvwrapper
After installation is complete always need a place to put virtualenvs related files and configuration, right?
# Create a virtual environment management directory mkdir ~ / .virtualenvs # Add the virtual environment management directory to the current user's environment variables, in order to facilitate future directly into # .Bashrc directly added at the following two lines, saving, and Source ~ / .bashrc or re-open a terminal export WORKON_HOME=$HOME/.virtualenvs source /usr/local/bin/virtualenvwrapper.sh
3, to create a virtual environment
# Python to create the specified version of the virtual environment
mkvirtualenv -p / usr / bin / python # tf2_cpu attention python version, some computers will be python3
4, using
# Activate the virtual environment, the user name in front of the terminal after activation is successful there will be a (tf2_cpu) logo, that you present in a virtual environment
workon tf2_cpu
# Exit virtual environment
deactivate
# Delete a virtual environment, which I did not try
rmvirtualenv tf2_cpu
5, installed in the virtual environment dependencies
First, enter the virtual environment, followed by the installation:
pip install tensorflow==2.0.0-alpha0 -i https://pypi.douban.com/simple pip install opencv-python -i https://pypi.douban.com/simple pip install pandas -i https://pypi.douban.com/simple
# Be careful here, after the installation finished import tensorflow the error. This is supposed to do? It was followed by Ann
PIP install TF-Nightly -i HTTPS: // pypi.douban.com/simple # where there is error has reminded, operating through a bunch of packages installed, I do not quite understand, but not being given the import tensorflow PIP install PANDAS # watercress source error, the error is not directly mounted, this is what action? Python -m PIP install matplotlib # installation matplotlib, I do not know why this command and the other looks a little bit different
Then start your run. .