Deprecate ModelCardGenerator component

Co-authored-by: Suzen Fylke <codesue@users.noreply.github.com>
Co-authored-by: Hannes Hapke <hanneshapke@users.noreply.github.com>

Author: codesue

README.md

@@ -35,7 +35,13 @@ The Model Card Toolkit is hosted on [PyPI](https://pypi.org/project/model-card-t
 
 ## Model Card Generation on TFX
 
-If you are using [TensorFlow Extended (TFX)](https://www.tensorflow.org/tfx), you can incorporate model card generation into your TFX pipeline via the [`ModelCardGenerator`](https://www.tensorflow.org/responsible_ai/model_card_toolkit/api_docs/python/model_card_toolkit/ModelCardGenerator) component. See our [guide](model_card_toolkit/documentation/guide/templates.md) for more details, [this case study](model_card_toolkit/documentation/examples/MLMD_Model_Card_Toolkit_Demo.ipynb) for a demonstration.
+If you are using [TensorFlow Extended (TFX)](https://www.tensorflow.org/tfx), you can
+incorporate model card generation into your TFX pipeline via the `ModelCardGenerator`
+component. The `ModelCardGenerator` component is moving to the
+[TFX Addons](https://github.com/tensorflow/tfx-addons) library and will not be
+included in Model Card Toolkit from version 1.4.0. This page will be updated to
+include the new links for the Model Cards in TFX
+guide and the end-to-end demo when that migration is completed.
 
 ## Schema
 

model_card_toolkit/__init__.py

@@ -14,6 +14,5 @@
 """A module to streamline and automate generation of Model Cards."""
 from model_card_toolkit.core import ModelCardToolkit
 from model_card_toolkit.model_card import *
-from model_card_toolkit.tfx.component import ModelCardGenerator
 from model_card_toolkit.utils.source import *
 from model_card_toolkit.version import __version__

model_card_toolkit/core_test.py

@@ -26,8 +26,11 @@
 from model_card_toolkit.utils import source as src
 from model_card_toolkit.utils.testdata import testdata_utils
 from model_card_toolkit.utils.testdata.tfxtest import TfxTest
+from model_card_toolkit.utils.tfx_util import _TFX_METRICS_TYPE
+from model_card_toolkit.utils.tfx_util import _TFX_STATS_TYPE
 import tensorflow_model_analysis as tfma
-from tfx.types import standard_artifacts
+
+from ml_metadata.proto import metadata_store_pb2
 
 
 class CoreTest(parameterized.TestCase, TfxTest):
@@ -115,11 +118,11 @@ def test_scaffold_assets_with_source(self, output_file_format: str,
       self._write_tfdv(tfdv_path, train_dataset_name, train_features,
                        eval_dataset_name, eval_features, mlmd_store)
       model_evaluation_artifacts = mlmd_store.get_artifacts_by_type(
-          standard_artifacts.ModelEvaluation.TYPE_NAME)
+          _TFX_METRICS_TYPE)
       example_statistics_artifacts = mlmd_store.get_artifacts_by_type(
-          standard_artifacts.ExampleStatistics.TYPE_NAME)
-      pushed_model_artifact = standard_artifacts.PushedModel()
-      pushed_model_artifact.uri = pushed_model_path
+          _TFX_STATS_TYPE)
+      # Use placeholder artifact to avoid introducing tfx as a dependency
+      pushed_model_artifact = metadata_store_pb2.Artifact(uri=pushed_model_path)
       tfma_src = src.TfmaSource(
           model_evaluation_artifacts=model_evaluation_artifacts,
           metrics_exclude=['average_loss'])

model_card_toolkit/documentation/README.md

@@ -33,16 +33,12 @@ html = mct.export_format()
 
 *   [Standalone Model Card Toolkit](https://colab.sandbox.google.com/github/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/Standalone_Model_Card_Toolkit_Demo.ipynb)
 *   [Scikit-Learn with Model Card Toolkit](https://colab.sandbox.google.com/github/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/Scikit_Learn_Model_Card_Toolkit_Demo.ipynb)
-*   [MLMD with Model Card Toolkit](https://colab.sandbox.google.com/github/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/MLMD_Model_Card_Toolkit_Demo.ipynb)
 
 If you want to try out the Model Card Toolkit (MCT) right away, you can run the
 [standalone Model Card Toolkit demo](https://colab.sandbox.google.com/github/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/Standalone_Model_Card_Toolkit_Demo.ipynb),
 or the
 [Scikit-Learn Model Card Toolkit demo](https://colab.sandbox.google.com/github/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/Scikit_Learn_Model_Card_Toolkit_Demo.ipynb),
-which demonstrates how MCT can be used in a Scikit-Learn workspace. If you are
-using [MLMD/TFX](https://www.tensorflow.org/tfx) and want to incorporate MCT
-into your workflow, you can try out the
-[MLMD Model Card Toolkit demo](https://colab.sandbox.google.com/github/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/MLMD_Model_Card_Toolkit_Demo.ipynb).
+which demonstrates how MCT can be used in a Scikit-Learn workspace.
 
 These demos can be run directly from your browser. Click
 [here](https://github.com/tensorflow/model-card-toolkit/blob/master/model_card_toolkit/documentation/examples/README.md)