首先是完善的不报错的代码及输出：

# -*- coding: UTF-8 -*-
import sys
reload(sys)
sys.setdefaultencoding('utf8')

from tensorflow.contrib.framework import arg_scope,add_arg_scope

def print_hh(name = 'ss'):
    print(name)


@add_arg_scope
def print_ee(name,add_args):
    print(name)
    print(add_args)

add_arg_scope(print_hh)


with arg_scope([print_hh],name='xx'):
    print_hh()
    print_hh('yy')


with arg_scope([print_ee],add_args='pp'):
    with arg_scope([print_ee], add_args='mm'):
        print_ee('test1')
    print_ee('test2')
    
输出：
ss
yy
test1
mm
test2
pp

从中可以看到arg_scope的作用范围，也可以看出其与函数参数默认值之间的优先权，但是问题就在于@add_arg_scope与add_arg_scope()函数的作用：

@add_arg_scope
def print_ee(name,add_args,add_p = 'learn'):
    print(name)
    print(add_args)
    print(add_p)

with arg_scope([print_ee],add_args='pp'):
    with arg_scope([print_ee], add_args='mm'):
        print_ee('test1')
    print_ee('test2')

with arg_scope([print_ee],add_args='pp',add_p='cc'):
    print_ee('test3')
    
    
输出：
test1
mm
learn
test2
pp
learn
test3
pp
cc

def print_hh(name = 'ss'):
    print(name)
add_arg_scope(print_hh)

with arg_scope([print_hh],name='xx'):
    print_hh()
    print_hh('yy')
    
输出：
ss
yy

可以看出add_arg_scope()与@add_arg_scope在优先权上是存在区别的，在add_arg_scope()中，在arg_scope()定义的参数优先权是不如函数的参数设置的默认值的.而在@add_arg_scope中，arg_scope()定义的参数优先权是大于函数参数的默认值的．

@add_arg_scope
def print_ee(para,name='cc'):
    print(para)
    print(name)
with arg_scope([print_ee],para='dd'):
    print_ee()
    print('oo')
    

输出：
dd
cc
oo

def print_hh(para,name = 'ss'):
    print(para)
    print(name)
add_arg_scope(print_hh)

with arg_scope([print_hh],para='xx'):
    print_hh()
    print_hh('yy')
    
输出：
Traceback (most recent call last):
  File "/home/saber/桌面/testForPRN/test_arg_scope.py", line 46, in <module>
    print_hh()
TypeError: print_hh() takes at least 1 argument (0 given)

可以看到add_arg_scope()与@add_arg_scope，前者并没有真正把参数传过去，而后者则是可以的，这让我怀疑add_arg_scope()函数是否有发挥作用，但是当我删除该函数时，在with arg_scope()函数内使用print_hh()函数却是会报错，这说明它的确发挥这自己的作用．

并且

@add_arg_scope
def print_ee(add_p = 'learn'):
    print(add_p)

with arg_scope([print_ee],add_p='pp'):
    print_ee()

输出：

pp

@add_arg_scope
def print_ee(add_p):
    print(add_p)

with arg_scope([print_ee],add_p='pp'):
    print_ee('d')
    
    
输出：
Traceback (most recent call last):
  File "/home/saber/桌面/testForPRN/test_arg_scope.py", line 55, in <module>
    print_ee('s')
  File "/usr/local/lib/python2.7/dist-packages/tensorflow/contrib/framework/python/ops/arg_scope.py", line 182, in func_with_args
    return func(*args, **current_args)
TypeError: print_ee() got multiple values for keyword argument 'add_p'

可以看到@add_arg_scope传入的值引起了冲突,而add_arg_scope()就不存在这个问题

def print_hh(add_p):
    print(add_p)
add_arg_scope(print_hh)

with arg_scope([print_hh],add_p='pp'):
    print_hh('d')
    
输出：
d

最后,只在代码中看到了这个，表示我的头还是很大

def add_arg_scope(func):
  """Decorates a function with args so it can be used within an arg_scope.
  Args:
    func: function to decorate.
  Returns:
    A tuple with the decorated function func_with_args().
  """

def arg_scope(list_ops_or_scope, **kwargs):
  """Stores the default arguments for the given set of list_ops.
  For usage, please see examples at top of the file.
  Args:
    list_ops_or_scope: List or tuple of operations to set argument scope for or
      a dictionary containing the current scope. When list_ops_or_scope is a
      dict, kwargs must be empty. When list_ops_or_scope is a list or tuple,
      then every op in it need to be decorated with @add_arg_scope to work.
    **kwargs: keyword=value that will define the defaults for each op in
              list_ops. All the ops need to accept the given set of arguments.
．．．
  Example of how to use tf.contrib.framework.add_arg_scope to enable your
  function to be called within an arg_scope later:
  @tf.contrib.framework.add_arg_scope
  def conv2d(*args, **kwargs)

参考资料