diff --git a/site/en/tutorials/keras/regression.ipynb b/site/en/tutorials/keras/regression.ipynb
index e3af85a7b9..ef6e1f5475 100644
--- a/site/en/tutorials/keras/regression.ipynb
+++ b/site/en/tutorials/keras/regression.ipynb
@@ -449,7 +449,7 @@
       },
       "outputs": [],
       "source": [
-        "normalizer.adapt(np.array(train_features))"
+        "normalizer.adapt(np.array(train_features).astype('float32'))"
       ]
     },
     {
@@ -494,7 +494,7 @@
         "with np.printoptions(precision=2, suppress=True):\n",
         "  print('First example:', first)\n",
         "  print()\n",
-        "  print('Normalized:', normalizer(first).numpy())"
+        "  print('Normalized:', normalizer(first.astype('float32')).numpy())"
       ]
     },
     {
@@ -545,7 +545,7 @@
       },
       "outputs": [],
       "source": [
-        "horsepower = np.array(train_features['Horsepower'])\n",
+        "horsepower = np.array(train_features['Horsepower'].astype('int'))\n",
         "\n",
         "horsepower_normalizer = layers.Normalization(input_shape=[1,], axis=None)\n",
         "horsepower_normalizer.adapt(horsepower)"
@@ -822,7 +822,7 @@
       },
       "outputs": [],
       "source": [
-        "linear_model.predict(train_features[:10])"
+        "linear_model.predict(train_features[:10].astype('int'))"
       ]
     },
     {
@@ -877,8 +877,8 @@
       "source": [
         "%%time\n",
         "history = linear_model.fit(\n",
-        "    train_features,\n",
-        "    train_labels,\n",
+        "    train_features.astype('float32'),\n",
+        "    train_labels.astype('float32'),\n",
         "    epochs=100,\n",
         "    # Suppress logging.\n",
         "    verbose=0,\n",
@@ -924,7 +924,7 @@
       "outputs": [],
       "source": [
         "test_results['linear_model'] = linear_model.evaluate(\n",
-        "    test_features, test_labels, verbose=0)"
+        "    test_features.astype('float32'), test_labels.astype('float32'), verbose=0)"
       ]
     },
     {
@@ -1173,8 +1173,8 @@
       "source": [
         "%%time\n",
         "history = dnn_model.fit(\n",
-        "    train_features,\n",
-        "    train_labels,\n",
+        "    train_features.astype('float32'),\n",
+        "    train_labels.astype('float32'),\n",
         "    validation_split=0.2,\n",
         "    verbose=0, epochs=100)"
       ]
@@ -1207,7 +1207,7 @@
       },
       "outputs": [],
       "source": [
-        "test_results['dnn_model'] = dnn_model.evaluate(test_features, test_labels, verbose=0)"
+        "test_results['dnn_model'] = dnn_model.evaluate(test_features.astype('float32'), test_labels.astype('float32'), verbose=0)"
       ]
     },
     {
@@ -1267,10 +1267,10 @@
       },
       "outputs": [],
       "source": [
-        "test_predictions = dnn_model.predict(test_features).flatten()\n",
+        "test_predictions = dnn_model.predict(test_features.astype('float32')).flatten()\n",
         "\n",
         "a = plt.axes(aspect='equal')\n",
-        "plt.scatter(test_labels, test_predictions)\n",
+        "plt.scatter(test_labels.astype('float32'), test_predictions.astype('float32'))\n",
         "plt.xlabel('True Values [MPG]')\n",
         "plt.ylabel('Predictions [MPG]')\n",
         "lims = [0, 50]\n",
@@ -1321,7 +1321,7 @@
       },
       "outputs": [],
       "source": [
-        "dnn_model.save('dnn_model.keras')"
+        "dnn_model.save('dnn_model.tf', save_format='tf')"
       ]
     },
     {
@@ -1341,10 +1341,10 @@
       },
       "outputs": [],
       "source": [
-        "reloaded = tf.keras.models.load_model('dnn_model.keras')\n",
+        "reloaded = tf.keras.models.load_model('dnn_model.tf')\n",
         "\n",
         "test_results['reloaded'] = reloaded.evaluate(\n",
-        "    test_features, test_labels, verbose=0)"
+        "    test_features.astype('float32'), test_labels.astype('float32'), verbose=0)"
       ]
     },
     {