A function which should obey the following signature:
Args: can only have following four arguments in any order:
predictions: Predictions Tensor or dict of Tensor created by given
estimator.
features: Input dict of Tensor objects created by input_fn which
is given to estimator.evaluate as an argument.
labels: Labels Tensor or dict of Tensor created by input_fn
which is given to estimator.evaluate as an argument.
config: config attribute of the estimator.
Returns: Dict of metric results keyed by name. Final metrics are a
union of this and estimator's existing metrics. If there is a name
conflict between this and estimators existing metrics, this will
override the existing one. The values of the dict are the results of
calling a metric function, namely a (metric_tensor, update_op) tuple.