Created by: pengwangucla

I want to write new evaluator for my project.

I check this example that run evaluator during training:

    def event_handler(event):
        if isinstance(event, paddle.event.EndIteration):
            if event.batch_id % 1000 == 0:
                print "Pass %d, Batch %d, Cost %f, %s" % (
                    event.pass_id, event.batch_id, event.cost, event.metrics)

                with gzip.open('params.tar.gz', 'w') as f:
                    parameters.to_tar(f)

        elif isinstance(event, paddle.event.EndPass):
            result = trainer.test(reader=paddle.batch(
                paddle.dataset.mnist.test(), batch_size=128))
            print "Test with Pass %d, Cost %f, %s\n" % (
                event.pass_id, result.cost, result.metrics)
            lists.append((event.pass_id, result.cost,
                          result.metrics['classification_error_evaluator']))

I found it is confusing that the evaluation method is select after result is returned. Is it correct that the method of evaluation should be selected in trainer.test(...) ? So that the network return the correct metric.

Can someone explain or better organize the arguments ?

@reyoung @luotao1 @lcy-seso