Support TensorFlow metric in TFOptimizer (intel#1593)

* test

* support stateless metric

* support metric in TFOptimizer

* fix tests

* fix style

* add tests

test_tf.py

@@ -84,33 +84,6 @@ def test_init_tfnet_from_session(self):
         self.assert_allclose(output_value, output_value_ref)
         self.assert_allclose(grad_input_value, grad_input_value_ref)
 
-    def test_tf_optimizer_with_sparse_gradient(self):
-        import tensorflow as tf
-
-        ids = np.random.randint(0, 10, size=[40])
-        labels = np.random.randint(0, 5, size=[40])
-        id_rdd = self.sc.parallelize(ids)
-        label_rdd = self.sc.parallelize(labels)
-        training_rdd = id_rdd.zip(label_rdd).map(lambda x: [x[0], x[1]])
-        with tf.Graph().as_default():
-            dataset = TFDataset.from_rdd(training_rdd,
-                                         names=["ids", "labels"],
-                                         shapes=[[], []],
-                                         types=[tf.int32, tf.int32],
-                                         batch_size=8)
-            id_tensor, label_tensor = dataset.tensors
-            embedding_table = tf.get_variable(
-                name="word_embedding",
-                shape=[10, 5])
-
-            embedding = tf.nn.embedding_lookup(embedding_table, id_tensor)
-            loss = tf.reduce_mean(tf.losses.
-                                  sparse_softmax_cross_entropy(logits=embedding,
-                                                               labels=label_tensor))
-            optimizer = TFOptimizer(loss, Adam(1e-3))
-            optimizer.optimize(end_trigger=MaxEpoch(1))
-            optimizer.sess.close()
-
     def test_tf_net_predict(self):
         resource_path = os.path.join(os.path.split(__file__)[0], "../../resources")
         tfnet_path = os.path.join(resource_path, "tfnet")