Created by: pengwangucla

A typical training steps in paddle is create parameter and then run the trainer:

    parameters = paddle.parameters.create(cost)
    trainer = paddle.trainer.SGD(cost=cost,
                                 parameters=parameters,
                                 update_equation=momentum_optimizer)

The topology is called twice in both create and SGD, why you need to do this, can you set a global variable of topo that know the topology has been set ?

@jacquesqiao