diff --git a/README.md b/README.md
index e394d123..be2d2f0f 100644
--- a/README.md
+++ b/README.md
@@ -131,7 +131,7 @@ using static Tensorflow.KerasApi;
 using Tensorflow;
 using Tensorflow.NumPy;
 
-var layers = new LayersApi();
+var layers = keras.layers;
 // input layer
 var inputs = keras.Input(shape: (32, 32, 3), name: "img");
 // convolutional layer
@@ -155,17 +155,19 @@ var model = keras.Model(inputs, outputs, name: "toy_resnet");
 model.summary();
 // compile keras model in tensorflow static graph
 model.compile(optimizer: keras.optimizers.RMSprop(1e-3f),
-    loss: keras.losses.CategoricalCrossentropy(from_logits: true),
+    loss: keras.losses.SparseCategoricalCrossentropy(from_logits: true),
     metrics: new[] { "acc" });
 // prepare dataset
 var ((x_train, y_train), (x_test, y_test)) = keras.datasets.cifar10.load_data();
+// normalize the input
 x_train = x_train / 255.0f;
-y_train = np_utils.to_categorical(y_train, 10);
 // training
 model.fit(x_train[new Slice(0, 2000)], y_train[new Slice(0, 2000)],
-          batch_size: 64,
-          epochs: 10,
-          validation_split: 0.2f);
+            batch_size: 64,
+            epochs: 10,
+            validation_split: 0.2f);
+// save the model
+model.save("./toy_resnet_model");
 ```
 
 The F# example for linear regression is available [here](docs/Example-fsharp.md).
diff --git a/docs/README-CN.md b/docs/README-CN.md
index 6fcb5195..47f19482 100644
--- a/docs/README-CN.md
+++ b/docs/README-CN.md
@@ -130,7 +130,7 @@ using static Tensorflow.KerasApi;
 using Tensorflow;
 using Tensorflow.NumPy;
 
-var layers = new LayersApi();
+var layers = keras.layers;
 // input layer
 var inputs = keras.Input(shape: (32, 32, 3), name: "img");
 // convolutional layer
@@ -154,17 +154,19 @@ var model = keras.Model(inputs, outputs, name: "toy_resnet");
 model.summary();
 // compile keras model in tensorflow static graph
 model.compile(optimizer: keras.optimizers.RMSprop(1e-3f),
-    loss: keras.losses.CategoricalCrossentropy(from_logits: true),
+    loss: keras.losses.SparseCategoricalCrossentropy(from_logits: true),
     metrics: new[] { "acc" });
 // prepare dataset
 var ((x_train, y_train), (x_test, y_test)) = keras.datasets.cifar10.load_data();
+// normalize the input
 x_train = x_train / 255.0f;
-y_train = np_utils.to_categorical(y_train, 10);
 // training
 model.fit(x_train[new Slice(0, 2000)], y_train[new Slice(0, 2000)],
-          batch_size: 64,
-          epochs: 10,
-          validation_split: 0.2f);
+            batch_size: 64,
+            epochs: 10,
+            validation_split: 0.2f);
+// save the model
+model.save("./toy_resnet_model");
 ```
 
 此外，Tensorflow.NET也支持用F#搭建上述模型进行训练和推理。