服务器下装tensorflow和caffe

Anaconda

1、bash xxx.sh

-p PATH -u

2、激活Anaconda的环境变量

export PATH=”/data/lzy/Anaconda3/bin:$PATH”

3、更换清华源

conda config –add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/

conda config –add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/

conda config –set show_channel_urls yes

4、创建TensorFlow的环境

conda create -n tensorflow python=3.6
- 激活环境
source activate tensorflow
- 退出环境
source deactivate tensorflow

5、显示环境中已经安装的包

conda list

6、安装TensorFlow GPU

pip install –upgrade /data/lzy/tensorflow_gpu-1.5.0-cp36-cp36m-manylinux1_x86_64.whl 【这个版本高了】
pip install –upgrade /data/lzy/tensorflow_gpu-1.2.0-cp36-cp36m-manylinux1_x86_64.whl 【换个低版本的】

Caffe

1、unzip xxx.zip -d PATH

2、http://douglasncode.blogspot.com/2017/03/install-caffe-with-python36-in-anaconda.html

https://www.cnblogs.com/go-better/p/7161006.html

3、一些临时环境变量

export PATH=”/data/lzy/Anaconda3/bin:$PATH” export PYTHONPATH=/data/lzy/caffe/python/:$PYTHONPATH

4、安装requirements.txt文件中需要的包

for req in $(cat requirements.txt); do pip install$req; done
如果加速：
for req in $(cat requirements.txt); do pip install -i https://pypi.tuna.tsinghua.edu.cn/simple

$req; done

pip加速

pip install -i https://pypi.tuna.tsinghua.edu.cn/simple

packagename