Ubuntu16.04 安装TensorFlow1.4.1

1、安装anaconda 参考https://blog.csdn.net/lwplwf/article/details/79162470

2、使用anaconda创建一个虚拟环境 

conda create -n <环境名称> python=3.5(python版本)

3、激活上一步骤创建的虚拟环境

#
# To activate this environment, use:
# > source activate tf1.4
#
# To deactivate this environment, use:
# > source deactivate tf1.4
#

4、用anaconda安装最新的TensorFlow版本

参考https://blog.csdn.net/qq_35203425/article/details/79965389

       https://blog.csdn.net/u013550000/article/details/80042471

5、激活环境之后测试

import tensorflow as tf

>>> tf.__version__
'1.4.1'

6、测试

 import tensorflow as tf
 hello = tf.constant('hello,tensorflow')
 sess = tf.Session()
 print(sess.run(hello))

测试时候的一些警告：

>>> sess = tf.Session()
2019-09-02 11:27:36.323811: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA

原因：安装的是cpu版本（后续还会安装GPU版本）

解决：在代码中加入以下代码，忽略警告

import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'

 

参考：https://blog.csdn.net/hq86937375/article/details/79696023

7、测试结果

(tf1.4) s@s:~$ python
Python 3.5.4 |Continuum Analytics, Inc.| (default, Aug 14 2017, 13:26:58) 
[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> import os 
>>> os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
>>> hello = tf.constant('hello,tensorflow')
>>> sess = tf.Session()
>>> print(sess.run(hello))
b'hello,tensorflow'
>>> 

ps:新建一个py文件

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
'''
import tensorflow as tf
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
hello = tf.constant('hello,tensorflow')
sess = tf.Session()
print(sess.run(hello))
'''

import tensorflow as tf
import numpy as np

#去除cpu版本的警告
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'

#creat data
x_data = np.random.rand(100).astype(np.float32)     ##输入值[0，1)之间的随机数
y_data = x_data * 0.1 + 0.3     ##预测值

###creat tensorflow structure strat###
# 构造要拟合的线性模型
Weights = tf.Variable(tf.random_uniform([1],-1.0,1.0))    
biases = tf.Variable(tf.zeros([1]))     
y = Weights * x_data + biases   

# 定义损失函数和训练方法
loss = tf.reduce_mean(tf.square(y-y_data))   ##最小化方差    
optimizer = tf.train.GradientDescentOptimizer(0.5)
train = optimizer.minimize(loss)

# 初始化变量
'''
init = tf.initialize_all_variables()
'''
init = tf.global_variables_initializer()
###creat tensorflow structure end###

# 启动
sess = tf.Session()
sess.run(init)

# 训练拟合，每一步训练队Weights和biases进行更新
count = 1
for step in range(201):
    sess.run(train)
    if step % 20 == 0:  
        print(step,sess.run(Weights),sess.run(biases)) 

# 得到最优拟合结果 W接近于0.1，b接近于0.3

在conda空间tf1.4中直接运行./test1.py，迭代200次后的结果如下

s@s:~$ source activate tf1.4
(tf1.4) s@s:~$ cd TF_study/
(tf1.4) s@s:~/TF_study$ ./test1.py 
0 [0.36548096] [0.21200436]
20 [0.15784316] [0.2683353]
40 [0.11386405] [0.2924105]
60 [0.10332299] [0.29818094]
80 [0.10079648] [0.299564]
100 [0.10019091] [0.2998955]
120 [0.10004577] [0.29997495]
140 [0.10001099] [0.299994]
160 [0.10000264] [0.29999858]
180 [0.10000062] [0.29999968]
200 [0.10000015] [0.29999992]