为什么我们使用 tf.name_scope() [英] Why do we use tf.name_scope()

问题描述

我一直在阅读他们编写的关于 TensorFlow 的教程

I've been reading the tutorials on TensorFlow where they have written

with tf.name_scope('read_inputs') as scope:
    # something

例子

a = tf.constant(5)

和

with tf.name_scope('s1') as scope:
    a = tf.constant(5)

似乎有同样的效果.那么，我们为什么要使用 name_scope?

seem to have the same effect. So, why do we use name_scope?

推荐答案

我没有看到重用常量的用例，但这里有一些关于作用域和变量共享的相关信息.

I don't see the use case for reusing constants but here is some relevant information on scopes and variable sharing.

范围

• name_scope 将范围作为前缀添加到所有操作

• name_scope will add scope as a prefix to all operations

variable_scope 将范围作为前缀添加到所有变量和操作

variable_scope will add scope as a prefix to all variables and operations

实例化变量

• tf.Variable() 构造器使用当前的 name_scope 和 variable_scope

• tf.Variable() constructer prefixes variable name with current name_scope and variable_scope

tf.get_variable() 构造函数忽略 name_scope 并且只用当前的 variable_scope

tf.get_variable() constructor ignores name_scope and only prefixes name with the current variable_scope

例如:

with tf.variable_scope("variable_scope"):
     with tf.name_scope("name_scope"):
         var1 = tf.get_variable("var1", [1])

with tf.variable_scope("variable_scope"):
     with tf.name_scope("name_scope"):
         var2 = tf.Variable([1], name="var2")

生产

var1 = <tf.Variable 'variable_scope/var1:0' shape=(1,) dtype=float32_ref>

var2 = <tf.Variable 'variable_scope/name_scope/var2:0' shape=(1,) dtype=string_ref>

重用变量

• 总是使用tf.variable_scope来定义共享变量的范围

重用变量的最简单方法是使用reuse_variables()，如下所示

The easiest way to do reuse variables is to use the reuse_variables() as shown below

with tf.variable_scope("scope"):
    var1 = tf.get_variable("variable1",[1])
    tf.get_variable_scope().reuse_variables()
    var2=tf.get_variable("variable1",[1])
assert var1 == var2

• tf.Variable() 总是创建一个新变量，当一个变量用一个已经用过的名字构造时，它只是附加 _1, _2> 等等 - 这可能会导致冲突:(
• tf.Variable() always creates a new variable, when a variable is constructed with an already used name it just appends _1, _2 etc. to it - which can cause conflicts :(

这篇关于为什么我们使用 tf.name_scope()的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！