mac上默认装有python，先看一下python的版本，
image.png
python3 以上的版本默认有pip ，所以2.7需要手动安装pip，安装命令
sudo easy_install pip
image.png
image.png
tensorflow1.2以上版本macos上不支持gpu，现在tensorflow1.8已经发布了。
安装cpu版本：
pip install tensorflow
image.png
失败了：
Collecting html5lib==0.9999999 (from tensorboard<1.9.0,>=1.8.0->tensorflow)
Using cached https://files.pythonhosted.org/packages/ae/ae/bcb60402c60932b32dfaf19bb53870b29eda2cd17551ba5639219fb5ebf9/html5lib-0.9999999.tar.gz
matplotlib 1.3.1 requires nose, which is not installed.
matplotlib 1.3.1 requires tornado, which is not installed.
Installing collected packages: numpy, pbr, funcsigs, six, mock, enum34, wheel, astor, backports.weakref, termcolor, gast, protobuf, futures, grpcio, absl-py, markdown, html5lib, bleach, werkzeug, tensorboard, tensorflow
Found existing installation: numpy 1.8.0rc1
Cannot uninstall 'numpy'. It is a distutils installed project and thus we cannot accurately determine which files belong to it which would lead to only a partial uninstall.
sniper:~ chris$ cd /usr/lib/python2.7/
貌似少一些依赖关系、还有不兼容的问题
先执行两命令看看
sudo easy_install nose
sudo easy_install tornado
image.png
再解决兼容问题：
在终端运行：
sudo pip install --ignore-installed numpy
image.png
完成后运行
pip install tensorflow
image.png
执行
pip install --user tensorflow
image.png
按照错误提示命令执行,但是依然会报错 ,原因是没有权限.
所以执行:
sudo pip install --upgrade pip
image.png
最后再执行
pip install --user tensorflow
验证一下:
import tensorflow as tf
hello=tf.constant("hello tensorflow")
sess=tf.Session()
print(sess.run(hello))
没毛病
image.png