Rename keras models function name

sqlflow_models/deep_embedding_cluster.py

@@ -106,10 +106,10 @@ def __init__(self,
 
         self.clustering_layer = ClusteringLayer(name='clustering', n_clusters=self._n_clusters)
 
-    def default_optimizer(self):
+    def optimizer(self):
         return self._cluster_optimizer
 
-    def default_loss(self):
+    def loss(self):
         return self._default_loss
 
     @staticmethod

sqlflow_models/dnnclassifier.py

@@ -25,18 +25,14 @@ def call(self, inputs):
             x = hidden_layer(x)
         return self.prediction_layer(x)
 
-    def default_optimizer(self):
+    def optimizer(self):
         """Default optimizer name. Used in model.compile."""
         return tf.keras.optimizers.Adagrad(lr=0.1)
 
-    def default_loss(self):
+    def loss(self):
         """Default loss function. Used in model.compile."""
         return 'sparse_categorical_crossentropy'
 
-    def default_training_epochs(self):
-        """Default training epochs. Used in model.fit."""
-        return 2
-
     def prepare_prediction_column(self, prediction):
         """Return the class label of highest probability."""
         return prediction.argmax(axis=-1)

sqlflow_models/lstmclassifier.py

@@ -27,10 +27,10 @@ def __init__(self, feature_columns, stack_units=[32], hidden_size=64, n_classes=
         if self.n_classes == 2:
             # special setup for binary classification
             pred_act = 'sigmoid'
-            self.loss = 'binary_crossentropy'
+            self._loss = 'binary_crossentropy'
         else:
             pred_act = 'softmax'
-            self.loss = 'categorical_crossentropy'
+            self._loss = 'categorical_crossentropy'
         self.pred = tf.keras.layers.Dense(n_classes, activation=pred_act)
 
     def call(self, inputs):
@@ -43,17 +43,14 @@ def call(self, inputs):
         x = self.hidden(x)
         return self.pred(x)
 
-    def default_optimizer(self):
+    def optimizer(self):
         """Default optimizer name. Used in model.compile."""
         return 'adam'
 
-    def default_loss(self):
+    def loss(self):
         """Default loss function. Used in model.compile."""
-        return self.loss
+        return self._loss
 
-    def default_training_epochs(self):
-        """Default training epochs. Used in model.fit."""
-        return 1
 
     def prepare_prediction_column(self, prediction):
         """Return the class label of highest probability."""

tests/base.py

@@ -22,11 +22,11 @@ def setUp(self):
         def test_train_and_predict(self):
             self.setUp()
 
-            self.model.compile(optimizer=self.model.default_optimizer(),
-                loss=self.model.default_loss(),
+            self.model.compile(optimizer=self.model.optimizer(),
+                loss=self.model.loss(),
                 metrics=["accuracy"])
             self.model.fit(train_input_fn(self.features, self.label),
-                epochs=self.model.default_training_epochs(),
+                epochs=1,
                 steps_per_epoch=100, verbose=0)
             loss, acc = self.model.evaluate(eval_input_fn(self.features, self.label))
             print(loss, acc)

tests/test_deep_embedding_cluster.py

@@ -90,8 +90,8 @@ def setUp(self):
     def test_train_and_predict(self):
         self.setUp()
 
-        self.model.compile(optimizer=self.model.default_optimizer(),
-                           loss=self.model.default_loss())
+        self.model.compile(optimizer=self.model.optimizer(),
+                           loss=self.model.loss())
         self.model.sqlflow_train_loop(train_input_fn(self.features, self.label))
         metric = evaluate(x=eval_input_fn(self.features, self.label), y=self.label, model=self.model)
         print(metric)