Use anaconda to create a new virtual environment and install GPU version of pytorch

Use anaconda to create a new virtual environment and install GPU version of pytorch
In the actual applications, we have to use different frameworks. In this case, it is better to install these frameworks in different virtual environments so as to aviod the interference among these frameworks.
Detailed steps are as follows:
(1) create an virtual environment [1] :
conda create -n pytorch1.1.0 python=3.6
(2) enter the environment :
source activate pytorch1.1.0
(3) watch the libraries that has been installed in this environment:
pip list
(4) run this code to install pytorch (note that cuda has been installed before on my computer) [2] :
conda install pytorch torchvision cudatoolkit=9.2 -c pytorch
(5) install some common libraires including opencv-python, sklearn, scikit-image, pandas etc. :
pip install opencv-python
pip install sklearn
pip install scikit-image
pip install pandas
(6) watch the version of pytorch (Important notes: The version of pytorch I installed is 1.1.0 and my python version is 3.6.2.)
python
import torch
(7) If you want to exit the virtual environment, then use the following code:
source deactivate pytorch1.1.0

Reference
[1] https://blog.csdn.net/sizhi_xht/article/details/80964099
[2] https://pytorch.org/