expand_dims to correctly test multi-hot

Without expanding dimensions, multi-hot encoded columns are not correctly illustrated (they will be parsed as single-record multi-label, not single-label multi-record).

(Another question is whether we need multi-hot for these columns instead of one-hot, as they are not multi-label?)

Author: Daniel Sparing

site/en/tutorials/structured_data/preprocessing_layers.ipynb

@@ -502,7 +502,7 @@
         "test_type_layer = get_category_encoding_layer(name='Type',\n",
         "                                              dataset=train_ds,\n",
         "                                              dtype='string')\n",
-        "test_type_layer(test_type_col)"
+        "test_type_layer(tf.expand_dims(test_type_col, -1))"
       ]
     },
     {
@@ -527,7 +527,7 @@
         "                                             dataset=train_ds,\n",
         "                                             dtype='int64',\n",
         "                                             max_tokens=5)\n",
-        "test_age_layer(test_age_col)"
+        "test_age_layer(tf.expand_dims(test_age_col, -1))"
       ]
     },
     {