Foreword
Before writing a few blog about TensorFlow 1.x GPU version installed, but almost never learned how. Prior to engage in basic things Machine Learning and Data Mining aspects, rarely used NN, although a few have seen the relevant code, but did not understand how too, static picture is somewhat complicated, difficult to like me for the chicken dish there is so little little point and orz. . .
Fortunately, however, the second half of last year TensorFlow 2.0 is finally out, and learned some time ago installed, the feeling you really want a lot simpler, probably because I use tensorflow.keras it. Anyway, on with a cool finish fix.
installation
I would say something about the installation process, do not know how much simpler than TensorFlow 1.x GPU version installed.
Above all, in the anaconda, so there is no anaconda would install a bar. The anaconda install on your stamp here
Then we want to set the mirror in Tsinghua University, seems quite a long time before this image was blocked, but now seems to be used, see details here
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/ conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/ conda config --set show_channel_urls yes
Next, start operation
conda create -n tf2. 0 python = 3.7
then
activate tf2.0 conda install tensorflow-gpu==2.0
Remember, this must be conda Install
Once the installation is complete, test test
Yes, success!
Then is pleasant Code Time
Section
Here on our TensorFlow 2.0 GPU installed, here for everyone to recommend a material , I feel pretty good, some time ago that looked at the study. I hope this tutorial can help to you, do not say, and then I go and change the code to run the experiment in the laboratory salted fish most of the year, winter and next semester is basically no what matter, can seriously mess things up research, 2333333.