Updating documentation:

* Switching to using `RecordBatchToExamplesEncoder`
* Utilizing `tft.DatasetMetadata.from_feature_spec`

PiperOrigin-RevId: 523723569

Author: zoyahav

docs/get_started.ipynb

@@ -422,12 +422,7 @@
         "\n",
         "The previous code examples used this format. The metadata contains the schema that defines the layout of the data and how it is read from and written to various formats. Even this in-memory format is not self-describing and requires the schema in order to be interpreted as tensors.\n",
         "\n",
-        "Again, here is the definition of the schema for the example data:\n",
-        "\n",
-        "\u003c!--\n",
-        "TODO(b/223384488): Switch to `tft.DatasetMetadata.from_feature_spec` once\n",
-        "version 1.8 is released.\n",
-        "--\u003e"
+        "Again, here is the definition of the schema for the example data:"
       ]
     },
     {
@@ -438,15 +433,13 @@
       },
       "outputs": [],
       "source": [
-        "from tensorflow_transform.tf_metadata import dataset_metadata\n",
-        "from tensorflow_transform.tf_metadata import schema_utils\n",
+        "import tensorflow_transform as tft\n",
         "\n",
-        "raw_data_metadata = dataset_metadata.DatasetMetadata(\n",
-        "      schema_utils.schema_from_feature_spec({\n",
+        "raw_data_metadata = tft.DatasetMetadata.from_feature_spec({\n",
         "        's': tf.io.FixedLenFeature([], tf.string),\n",
         "        'y': tf.io.FixedLenFeature([], tf.float32),\n",
         "        'x': tf.io.FixedLenFeature([], tf.float32),\n",
-        "    }))"
+        "    })"
       ]
     },
     {