TensorFlow教程（0）

先上代码:

import tensorflow as tf
import numpy as np
#created data
x_data=np.random.rand(100).astype(np.float32)
y_data=x_data*0.1+0.3

###create tensorflow structure start###
weights=tf.Variable(tf.random_uniform([1],-1.0,1.0))
biases=tf.Variable(tf.zeros([1]))

y=weights*x_data+biases

loss=tf.reduce_mean(tf.square(y-y_data))
optimizer=tf.train.GradientDescentOptimizer(0.5)
train=optimizer.minimize(loss)
init=tf.global_variables_initializer()
###create tensorflow structure end ###
sess=tf.Session()
sess.run(init) #very important 可以把run当做指针
for step in range(200):
    sess.run(train)
    if step%20==0:
    print(step,sess.run(weights),sess.run(biases))