TensorFlow的异常Reciprocal[T=DT_INT32](Variable_1/read)

在TF里，使用32位整数来计算倒数时会抛出这个异常：

Traceback (most recent call last):
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 1022, in _do_call
    return fn(*args)
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 1000, in _run_fn
    self._extend_graph()
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 1049, in _extend_graph
    self._session, graph_def.SerializeToString(), status)
  File "C:\python35\lib\contextlib.py", line 66, in __exit__
    next(self.gen)
  File "C:\python35\lib\site-packages\tensorflow\python\framework\errors_impl.py", line 469, in raise_exception_on_not_ok_status
    pywrap_tensorflow.TF_GetCode(status))
tensorflow.python.framework.errors_impl.InvalidArgumentError: No OpKernel was registered to support Op 'Reciprocal' with these attrs.  Registered devices: [CPU,GPU], Registered kernels:
  device='CPU'; T in [DT_FLOAT]
  device='CPU'; T in [DT_HALF]
  device='CPU'; T in [DT_DOUBLE]
  device='CPU'; T in [DT_COMPLEX64]
  device='CPU'; T in [DT_COMPLEX128]
  device='GPU'; T in [DT_FLOAT]
  device='GPU'; T in [DT_HALF]
  device='GPU'; T in [DT_DOUBLE]
  device='GPU'; T in [DT_INT64]


[[Node: Reciprocal = Reciprocal[T=DT_INT32](Variable_1/read)]]


During handling of the above exception, another exception occurred:


Traceback (most recent call last):
  File "D:\work\csdn\TF_API\src\TFAPI_6\TFAPI_6\TFAPI_17.py", line 31, in <module>
    demo.run_graph()
  File "D:\work\csdn\TF_API\src\TFAPI_6\TFAPI_6\TFAPI_17.py", line 21, in run_graph
    sess.run(init)
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 767, in run
    run_metadata_ptr)
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 965, in _run
    feed_dict_string, options, run_metadata)
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 1015, in _do_run
    target_list, options, run_metadata)
  File "C:\python35\lib\site-packages\tensorflow\python\client\session.py", line 1035, in _do_call
    raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.InvalidArgumentError: No OpKernel was registered to support Op 'Reciprocal' with these attrs.  Registered devices: [CPU,GPU], Registered kernels:
  device='CPU'; T in [DT_FLOAT]
  device='CPU'; T in [DT_HALF]
  device='CPU'; T in [DT_DOUBLE]
  device='CPU'; T in [DT_COMPLEX64]
  device='CPU'; T in [DT_COMPLEX128]
  device='GPU'; T in [DT_FLOAT]
  device='GPU'; T in [DT_HALF]
  device='GPU'; T in [DT_DOUBLE]
  device='GPU'; T in [DT_INT64]


[[Node: Reciprocal = Reciprocal[T=DT_INT32](Variable_1/read)]]


Caused by op 'Reciprocal', defined at:
  File "<string>", line 1, in <module>
  File "C:\python35\lib\idlelib\run.py", line 130, in main
    ret = method(*args, **kwargs)
  File "C:\python35\lib\idlelib\run.py", line 357, in runcode
    exec(code, self.locals)
  File "D:\work\csdn\TF_API\src\TFAPI_6\TFAPI_6\TFAPI_17.py", line 30, in <module>
    demo = DemoTF()
  File "D:\work\csdn\TF_API\src\TFAPI_6\TFAPI_6\TFAPI_17.py", line 16, in __init__
    self.reciprocal = tf.reciprocal(self.num2)
  File "C:\python35\lib\site-packages\tensorflow\python\ops\gen_math_ops.py", line 1964, in reciprocal
    result = _op_def_lib.apply_op("Reciprocal", x=x, name=name)
  File "C:\python35\lib\site-packages\tensorflow\python\framework\op_def_library.py", line 763, in apply_op
    op_def=op_def)
  File "C:\python35\lib\site-packages\tensorflow\python\framework\ops.py", line 2395, in create_op
    original_op=self._default_original_op, op_def=op_def)
  File "C:\python35\lib\site-packages\tensorflow\python\framework\ops.py", line 1264, in __init__
    self._traceback = _extract_stack()


InvalidArgumentError (see above for traceback): No OpKernel was registered to support Op 'Reciprocal' with these attrs.  Registered devices: [CPU,GPU], Registered kernels:
  device='CPU'; T in [DT_FLOAT]
  device='CPU'; T in [DT_HALF]
  device='CPU'; T in [DT_DOUBLE]
  device='CPU'; T in [DT_COMPLEX64]
  device='CPU'; T in [DT_COMPLEX128]
  device='GPU'; T in [DT_FLOAT]
  device='GPU'; T in [DT_HALF]
  device='GPU'; T in [DT_DOUBLE]
  device='GPU'; T in [DT_INT64]


[[Node: Reciprocal = Reciprocal[T=DT_INT32](Variable_1/read)]]

从异常信息上来看，它是不支持tf.int32的类型运算的，所以抛出异常。只支持下面的类型：

  device='CPU'; T in [DT_FLOAT]
  device='CPU'; T in [DT_HALF]
  device='CPU'; T in [DT_DOUBLE]
  device='CPU'; T in [DT_COMPLEX64]
  device='CPU'; T in [DT_COMPLEX128]
  device='GPU'; T in [DT_FLOAT]
  device='GPU'; T in [DT_HALF]
  device='GPU'; T in [DT_DOUBLE]
  device='GPU'; T in [DT_INT64]

代码如下：

#python 3.5.3/TensorFlow 1.0/win 10    
#2017-04-01 蔡军生  http://blog.csdn.net/caimouse    
#
#导入要使用的库
import tensorflow as tf
import numpy as np

#演示API的类
class DemoTF:
    def __init__(self):
        '''构造函数'''
        self.num1 = tf.Variable([1, -2, -3], dtype = tf.int32)
        self.num2 = tf.Variable([4, 5, -6], dtype = tf.int32)
        
        self.sign = tf.sign(self.num1)
        self.reciprocal = tf.reciprocal(self.num2)
        
    def run_graph(self):       
        init = tf.global_variables_initializer()  # 初始化变量
        with tf.Session() as sess:            
            sess.run(init)
            self.main(sess) #运行主函数
    def main(self, sess):
        '''主函数'''
        print(sess.run(self.sign)) 
        print(sess.run(self.reciprocal)) 
                
#主程序入口
if __name__ == "__main__":
    demo = DemoTF()
    demo.run_graph()

1. TensorFlow API攻略

3. C++标准模板库从入门到精通 

http://edu.csdn.net/course/detail/3324

4.跟老菜鸟学C++

5. 跟老菜鸟学python

6. 在VC2015里学会使用tinyxml库

7. 在Windows下SVN的版本管理与实战 

 http://edu.csdn.net/course/detail/2579