win10下配置tensorflow环境(anaconda+jupyter notebook)

降低python至3.5
conda install python=3.5

创建一个tensorflow 虚拟环境：
conda create -n tensorflow python=3.5

激活tensorflow虚拟环境
activate tensorflow

安装tensorflow
pip install tensorflow -i https://pypi.tuna.tsinghua.edu.cn/simple (从国内镜像源下载，速度极快且稳定)

然后测试
python

import tensorflow as tf
message=tf.constant(‘welcome’)
with tf.Sesiion() as sess:
print(sess.run(message).decode())

运行成功，但是jupyter notebook打开后输入
import tensorflow as tf
运行报错，需要在激活TensorFlow之后安装ipython和jupyter

(tensorflow)D:>conda install ipython
・・・
(tensorflow)D:>conda install jupyter
・・・
接着安装nb_conda包库
(tensorflow)D:>conda install nb_conda
然后启动jupyter notebook
结果显示

就成功了

文章来源: https://blog.csdn.net/m0_37723350/article/details/90238732