train_action

 

 

# 导入数值计算模块
import numpy as np
import tensorflow as tf

# 创建计算会话
sess = tf.Session()
# 生成数据，创建占位符和变量A
x_vales = np.random.normal(1, 0.1, 100)
y_vals = np.repeat(10., 100)
x_data = tf.placeholder(shape=[1], dtype=tf.float32)
y_target = tf.placeholder(shape=[1], dtype=tf.float32)
A = tf.Variable(tf.random_normal(shape=[1]))

# 增加乘法操作
my_output = tf.multiply(x_data, A)
# 增加L2正则损失函数
loss = tf.square(my_output - y_target)

# 在运行之前，需要初始化变量
#init = tf.initialize_all_tables()
init = tf.tables_initializer()
sess.run(init)

# 声明变量的优化器

# 学习率的选取
my_opt = tf.train.GradientDescentOptimizer(learning_rate=0.2)
train_step = my_opt.minimize(loss)

# 训练算法
for i in range(100):
    rand_index = np.random.choice(100)
    rand_x = [x_vales[rand_index]]
    rand_y = [y_vals[rand_index]]
    sess.run(train_step, feed_dict={x_data: rand_x, y_target: rand_y})
    if (i + 1) % 25 == 0:
        print('Step #' + str(i + 1) + 'A = ' + str(sess.run(A)))
        print('Loss = ' + str(sess.run(loss, feed_dict={x_data: rand_x, y_target: rand_y})))
#
d = 3

 

 

 

 

An Op that initializes all tables. Note that if there are not tables the returned Op is a NoOp.

 

 

特征列和层一样具有内部状态，因此通常需要将它们初始化

 

Feature columns can have internal state, like layers, so they often need to be initialized. Categorical columns use lookup tables internally and these require a separate initialization op, tf.tables_initializer.

 

var_init = tf.global_variables_initializer()
table_init = tf.tables_initializer()
sess = tf.Session()
sess.run((var_init, table_init))

Once the internal state has been initialized you can run inputs like any other tf.Tensor:

 

 

 

# 导入数值计算模块
import numpy as np
import tensorflow as tf

# 构建计算图
# 生成数据，创建占位符和变量A
x_vales = np.random.normal(1, 0.1, 100)
y_vals = np.repeat(10., 100)
x_data = tf.placeholder(shape=[1], dtype=tf.float32)
y_target = tf.placeholder(shape=[1], dtype=tf.float32)
A = tf.Variable(tf.random_normal(shape=[1]))

# 增加乘法操作
my_output = tf.multiply(x_data, A)
# 增加L2正则损失函数
loss = tf.square(my_output - y_target)

# 声明变量的优化器
# 学习率的选取
my_opt = tf.train.GradientDescentOptimizer(learning_rate=0.2)
train_step = my_opt.minimize(loss)

# 运行计算图

# 创建计算会话
sess = tf.Session()

# 内部状态初始化完成后，您就可以像运行任何其他 tf.Tensor 一样运行 inputs：
# 特征列和层一样具有内部状态，因此通常需要将它们初始化。分类列会在内部使用对照表，而这些表需要单独的初始化指令 tf.tables_initializer。

var_init = tf.global_variables_initializer()
table_init = tf.tables_initializer()

sess.run((var_init, table_init))

# 训练算法
for i in range(100):
    rand_index = np.random.choice(100)
    rand_x = [x_vales[rand_index]]
    rand_y = [y_vals[rand_index]]
    sess.run(train_step, feed_dict={x_data: rand_x, y_target: rand_y})
    if (i + 1) % 25 == 0:
        print('Step #' + str(i + 1) + 'A = ' + str(sess.run(A)))
        print('Loss = ' + str(sess.run(loss, feed_dict={x_data: rand_x, y_target: rand_y})))

 

 

 

2018-05-12 16:57:22.358693: I T:\src\github\tensorflow\tensorflow\core\platform\cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2
Step #25A = [10.501938]
Loss = [0.00203205]
Step #50A = [9.105795]
Loss = [0.2731857]
Step #75A = [9.097782]
Loss = [0.5107153]
Step #100A = [9.557248]
Loss = [0.00200771]