Created by: hxk11111

版本、环境信息： 1）PaddlePaddle版本：1.6.2 2）python：3.6.3

训练信息 1）单机单卡

total_loss = self.alpha * losses[0] + self.beta * losses[1] + self.gamma * losses[2]
logging.info("Shape of total_loss: {}".format(total_loss.shape))
inference_program = fluid.default_main_program().clone(for_test=True)
optimizer = fluid.optimizer.Adam(learning_rate=self.learning_rate)
optimizer.minimize(total_loss)

在minimize total loss时报错

I0106 22:17:22.031168 64214 op_desc.cc:685] CompileTime infer shape on sum
I0106 22:17:22.031216 64214 op_desc.cc:685] CompileTime infer shape on expand_grad
I0106 22:17:22.031261 64214 op_desc.cc:685] CompileTime infer shape on unsqueeze2_grad
I0106 22:17:22.031307 64214 op_desc.cc:685] CompileTime infer shape on gather_nd_grad
I0106 22:17:22.031350 64214 op_desc.cc:685] CompileTime infer shape on cast
Traceback (most recent call last):
  File "models/train.py", line 158, in <module>
    train()
  File "models/train.py", line 56, in train
    accuracies, inference_program = gnn_model.train_net()
  File "/root/paddle/GNN-Table-Parsing/models/gnn_model.py", line 349, in train_net
    optimizer.minimize(total_loss)
  File "<decorator-gen-36>", line 2, in minimize
  File "/opt/conda/lib/python3.6/site-packages/paddle/fluid/wrapped_decorator.py", line 25, in __impl__
    return wrapped_func(*args, **kwargs)
  File "/opt/conda/lib/python3.6/site-packages/paddle/fluid/dygraph/base.py", line 78, in __impl__
    return func(*args, **kwargs)
  File "/opt/conda/lib/python3.6/site-packages/paddle/fluid/optimizer.py", line 678, in minimize
    no_grad_set=no_grad_set)
  File "/opt/conda/lib/python3.6/site-packages/paddle/fluid/optimizer.py", line 551, in backward
    no_grad_set, callbacks)
  File "/opt/conda/lib/python3.6/site-packages/paddle/fluid/backward.py", line 1085, in append_backward
    _append_backward_vars_(root_block, fwd_op_num, grad_to_var, grad_info_map)
  File "/opt/conda/lib/python3.6/site-packages/paddle/fluid/backward.py", line 891, in _append_backward_vars_
    op_desc.infer_shape(block.desc)
paddle.fluid.core_avx.EnforceNotMet:

--------------------------------------------
C++ Call Stacks (More useful to developers):
--------------------------------------------
0   std::string paddle::platform::GetTraceBackString<char const*>(char const*&&, char const*, int)
1   paddle::platform::EnforceNotMet::EnforceNotMet(std::__exception_ptr::exception_ptr, char const*, int)
2   paddle::operators::CastOpInferShape::operator()(paddle::framework::InferShapeContext*) const
3   std::_Function_handler<void (paddle::framework::InferShapeContext*), paddle::framework::details::OpInfoFiller<paddle::operators::CastOpInferShape, (paddle::framework::details::OpInfoFillType)4>::operator()(char const*, paddle::framework::OpInfo*) const::{lambda(paddle::framework::InferShapeContext*)#1}>::_M_invoke(std::_Any_data const&, paddle::framework::InferShapeContext*)
4   paddle::framework::OpDesc::InferShape(paddle::framework::BlockDesc const&) const

----------------------
Error Message Summary:
----------------------
Error: The input of cast op must be set at (/paddle/paddle/fluid/operators/cast_op.cc:44)