Created by: csx008

I need to use paddle.evaluator.precision_recall evaluator in each training pass while default is classification_error. When trying to change cost layer's evaluator, like:

cost = paddle.layer.classification_cost(
    input=predict,
    label=lbl,
    evaluator=paddle.evaluator.precision_recall
)

I get AssertionError:

File "/home/work/MyWorkspace/PaddlePaddle/python27-gcc482/lib/python2.7/site-packages/paddle/v2/config_base.py", line 52, in wrapped
    out = f(*args, **xargs)
  File "/home/work/MyWorkspace/PaddlePaddle/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/default_decorators.py", line 53, in __wrapper__
    return func(*args, **kwargs)
  File "/home/work/MyWorkspace/PaddlePaddle/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/layers.py", line 415, in wrapper
    return method(*args, **kwargs)
  File "/home/work/MyWorkspace/PaddlePaddle/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/layers.py", line 4627, in classification_cost
    __add_evaluator__(each_evaluator)
  File "/home/work/MyWorkspace/PaddlePaddle/python27-gcc482/lib/python2.7/site-packages/paddle/trainer_config_helpers/layers.py", line 4614, in __add_evaluator__
    assert hasattr(e, 'is_evaluator')

And set evaluator to the default classification_error also get this eror. Is my way OK?