In [1]:

    

import tensorflow as tf
import numpy as np


    

In [2]:

    

node1 = tf.constant(3.0, tf.float32)
node2 = tf.constant(4.0)
print(node2, node2)


    

    




Tensor("Const_1:0", shape=(), dtype=float32) Tensor("Const_1:0", shape=(), dtype=float32)

In [4]:

    

sess = tf.Session()


    

In [6]:

    

print(sess.run([node1, node2]))


    

    




[3.0, 4.0]

In [7]:

    

node3 = tf.add(node1, node2)


    

In [15]:

    

print('node 3: ', node3, '\n', sess.run([node3]))


    

    




node 3:  Tensor("Add:0", shape=(), dtype=float32) 
 [7.0]

In [22]:

    

a = tf.placeholder(tf.float32)
b = tf.placeholder(tf.float32)
adder_node = tf.add(a,b)
sess.run(adder_node, {a:[3,2] ,b:[3,2]})


    

    
Out[22]:





array([ 6.,  4.], dtype=float32)

In [23]:

    

weight = tf.Variable(.3, tf.float32)
bias = tf.Variable(-.5, tf.float32)
input = tf.placeholder(tf.float32)


    

In [24]:

    

linear_model = input * weight + bias


    

In [27]:

    

init = tf.global_variables_initializer()


    

In [28]:

    

sess.run(init)


    

In [32]:

    

print(sess.run(linear_model, {input:[1,2,3,4]}))


    

    




[-0.19999999  0.10000002  0.40000004  0.70000005]

In [34]:

    

error = tf.placeholder(tf.float32)
squared_deltas = tf.square(linear_model - error)
loss = tf.reduce_sum(squared_deltas)
print(sess.run(loss, {input:[1,2,3,4], error:[-.1, .2, .3, .3]}))


    

    




0.19

In [35]:

    

fixWeight = tf.assign(weight, -1)
fixBias = tf.assign(bias, 1)


    

In [37]:

    

sess.run([fixWeight, fixBias])


    

    
Out[37]:





[-1.0, 1.0]

In [40]:

    

print(sess.run(loss, {input:[1,2,3,4], error:[-0,-1,-2,-3]}))


    

    




0.0

In [45]:

    

optimizer = tf.train.GradientDescentOptimizer(0.01)
train = optimizer.minimize(loss)


    

In [46]:

    

for i in range(1000):
    sess.run(train, {input:[1,2,3,4], error:[-0,-1,-2,-3]})


    

In [48]:

    

print(sess.run([weight, bias]))


    

    




[-1.0, 1.0]

In [ ]: