Created by: YingjingLu

版本paddle 1.5 CUDA 97

复现代码：https://github.com/YingjingLu/Paddle-Issues/blob/master/Issue2.py 去掉gaussian_normal可以正确反向

  File ".\Issue2.py", line 420, in <module>
    _, grad_list = optim.minimize( loss )
  File "C:\Users\StevenLu\AppData\Local\Programs\Python\Python37\lib\site-packages\paddle\fluid\optimizer.py", line 498, in minimize
    no_grad_set=no_grad_set)
  File "C:\Users\StevenLu\AppData\Local\Programs\Python\Python37\lib\site-packages\paddle\fluid\optimizer.py", line 403, in backward
    no_grad_set, callbacks)
  File "C:\Users\StevenLu\AppData\Local\Programs\Python\Python37\lib\site-packages\paddle\fluid\backward.py", line 511, in append_backward
    grad_to_var, callbacks)
  File "C:\Users\StevenLu\AppData\Local\Programs\Python\Python37\lib\site-packages\paddle\fluid\backward.py", line 297, in _append_backward_ops_
    op.desc, cpt.to_text(no_grad_dict[block.idx]), grad_sub_block_list)
paddle.fluid.core.EnforceNotMet: grad_op_maker_ should not be null
Operator GradOpMaker has not been registered. at [D:\1.4.1\paddle\paddle/fluid/framework/op_info.h:64]