TensorFlow中Session的用法¶
作者：凯鲁嘎吉 - 博客园 http://www.cnblogs.com/kailugaji/
所用版本：python3.5.2，tensorflow1.8.0，tensorboard1.8.0
In [1]:
# 简单实现
In [2]:
import tensorflow as tf
In [3]:
a = tf.add(2, 5)
In [4]:
b = tf.multiply(a, 3)
In [5]:
sess = tf.Session()
In [6]:
sess.run(b)
Out[6]:
In [7]:
sess.run([a, b])
Out[7]:
In [8]:
# feed_dict参数：覆盖数据流图中的张量对象值
In [9]:
reset
In [10]:
import tensorflow as tf
In [11]:
a = tf.add(2, 5)
In [12]:
b = tf.multiply(a, 3)
In [13]:
sess = tf.Session()
In [14]:
aa = {a: 15} # 将a值替换为15
In [15]:
sess.run(b, feed_dict=aa)
Out[15]:
In [16]:
sess.close()
In [17]:
# 输入与占位符
In [18]:
reset
In [19]:
import tensorflow as tf
In [20]:
import numpy as np
In [21]:
# 创建一个长度为2，数据类型为int32的占位向量
In [22]:
a = tf.placeholder(tf.int32, shape=[2], name="my_imput")
In [23]:
# 将该占位向量视为其他任意张量对象，加以使用
In [24]:
b = tf.reduce_prod(a, name="prod_b")
In [25]:
c = tf.reduce_sum(a, name="sum_c")
In [26]:
# 完成数据流图的定义
In [27]:
d = tf.add(b ,c, name="add_d")
In [28]:
sess = tf.Session()
In [29]:
# 创建一个将传给feed_dict参数的字典
In [30]:
# 键：‘a’，指向占位符输出张量对象的句柄
In [31]:
# 值：一个值为[5, 3]，类型为int32的向量
In [32]:
aa = {a: np.array([5, 3], dtype=np.int32)}
In [33]:
# 计算d值，将aa的“值”传给a
In [34]:
sess.run(d, feed_dict=aa)
Out[34]:
In [35]:
# 这里计算d值依赖于a的输出
In [36]:
sess.close()