tf.py_func()就是个wrapper而已

tf.py_func()
Given a python function func, which takes numpy arrays as its arguments and returns numpy arrays as its outputs, wrap this function as an operation in a TensorFlow graph.
tf.py_func()的作用是将以numpy.arrary作为参数，并返回numpy.arrary的python函数包装成TensorFlow计算图中的op。

import numpy as np
import tensorflow as tf

def func(a, b):
    return a+b, a-b

a = tf.placeholder(tf.float32,[2,2],'a')
b = tf.placeholder(tf.float32,[2,2],'b')

res1, res2 = tf.py_func(func,[a,b],[tf.float32,tf.float32])

with tf.Session() as sess:
    res1_,res2_ = sess.run([res1,res2], feed_dict={a:[[1,2],[3,4]],b:[[5,6],[7,8]]})
    print(res1_)
    print(res2_)

输出：

[[ 6.  8.]
 [10. 12.]]
[[-4. -4.]
 [-4. -4.]]