MCT 1.1 release: update TFX dependencies and update MLMD_Model_Card_T…

…oolkit_Demo

PiperOrigin-RevId: 394550241

RELEASE.md

@@ -1,11 +1,14 @@
 <!-- mdlint off(HEADERS_TOO_MANY_H1) -->
 
-# Current Version (Still in Development)
+# Release 1.1.0
 
 ## Major Features and Improvements
 
 ## Bug fixes and other changes
 
+* Update TFX compatibility to TFX 1.2.
+* Fix bug where all datasets from MLMD were being compressed into one model_card.Dataset object.
+
 ## Breaking changes and Deprecations
 
 # Release 1.0.0

model_card_toolkit/documentation/examples/MLMD_Model_Card_Toolkit_Demo.ipynb

@@ -102,7 +102,7 @@
       "outputs": [],
       "source": [
         "!pip install --upgrade pip==20.2\n",
-        "!pip install \"tfx==0.26.0\"\n",
+        "!pip install \"tfx==1.2.0\"\n",
         "!pip install model-card-toolkit"
       ]
     },
@@ -151,7 +151,6 @@
         "from tfx.components import CsvExampleGen\n",
         "from tfx.components import Evaluator\n",
         "from tfx.components import Pusher\n",
-        "from tfx.components import ResolverNode\n",
         "from tfx.components import SchemaGen\n",
         "from tfx.components import StatisticsGen\n",
         "from tfx.components import Trainer\n",
@@ -167,7 +166,6 @@
         "from tfx.types import Channel\n",
         "from tfx.types.standard_artifacts import Model\n",
         "from tfx.types.standard_artifacts import ModelBlessing\n",
-        "from tfx.utils.dsl_utils import external_input\n",
         "\n",
         "import ml_metadata as mlmd"
       ]
@@ -290,7 +288,7 @@
         "# `pipeline_root` and `metadata_connection_config` may be passed to\n",
         "# InteractiveContext. Calls to InteractiveContext are no-ops outside of the\n",
         "# notebook.\n",
-        "context = InteractiveContext()"
+        "context = InteractiveContext(pipeline_name=\"Census Income Classification Pipeline\")"
       ]
     },
     {
@@ -322,7 +320,7 @@
       },
       "outputs": [],
       "source": [
-        "example_gen = CsvExampleGen(input=external_input(_data_root))\n",
+        "example_gen = CsvExampleGen(input_base=_data_root)\n",
         "context.run(example_gen)"
       ]
     },
@@ -356,7 +354,7 @@
       "outputs": [],
       "source": [
         "# Get the URI of the output artifact representing the training examples, which is a directory\n",
-        "train_uri = os.path.join(example_gen.outputs['examples'].get()[0].uri, 'train')\n",
+        "train_uri = os.path.join(example_gen.outputs['examples'].get()[0].uri, 'Split-train')\n",
         "\n",
         "# Get the list of files in this directory (all compressed TFRecord files)\n",
         "tfrecord_filenames = [os.path.join(train_uri, name)\n",
@@ -1231,10 +1229,10 @@
         "# different collections. \n",
         "model_card.quantitative_analysis.graphics.collection = filter_graphs(\n",
         "    model_card.quantitative_analysis.graphics.collection, TARGET_EVAL_GRAPH_NAMES)\n",
-        "model_card.model_parameters.data.eval.graphics.collection = filter_graphs(\n",
-        "    model_card.model_parameters.data.eval.graphics.collection, TARGET_DATASET_GRAPH_NAMES)\n",
-        "model_card.model_parameters.data.train.graphics.collection = filter_graphs(\n",
-        "    model_card.model_parameters.data.train.graphics.collection, TARGET_DATASET_GRAPH_NAMES)"
+        "model_card.model_parameters.data[0].graphics.collection = filter_graphs(\n",
+        "   model_card.model_parameters.data[0].graphics.collection, TARGET_DATASET_GRAPH_NAMES)\n",
+        "model_card.model_parameters.data[1].graphics.collection = filter_graphs(\n",
+        "    model_card.model_parameters.data[1].graphics.collection, TARGET_DATASET_GRAPH_NAMES)"
       ]
     },
     {
@@ -1254,13 +1252,15 @@
       },
       "outputs": [],
       "source": [
-        "model_card.model_parameters.data.train.graphics.description = (\n",
+        "model_card.model_parameters.data[0].name = 'train_set'\n",
+        "model_card.model_parameters.data[0].graphics.description = (\n",
         "    'This section includes graphs displaying the class distribution for the '\n",
         "    '“Race” and “Sex” attributes in our training dataset. We chose to '\n",
         "    'show these graphs in particular because we felt it was important that '\n",
         "    'users see the class imbalance.'\n",
         ")\n",
-        "model_card.model_parameters.data.eval.graphics.description = (\n",
+        "model_card.model_parameters.data[1].name = 'eval_set'\n",
+        "model_card.model_parameters.data[1].graphics.description = (\n",
         "    'Like the training set, we provide graphs showing the class distribution '\n",
         "    'of the data we used to evaluate our model’s performance. '\n",
         ")\n",

model_card_toolkit/model_card_toolkit.py

@@ -173,8 +173,9 @@ def _scaffold_model_card(self) -> ModelCard:
           graphics.annotate_eval_result_plots(model_card, eval_result)
 
       for stats_artifact in stats_artifacts:
-        train_stats = tfx_util.read_stats_proto(stats_artifact.uri, 'train')
-        eval_stats = tfx_util.read_stats_proto(stats_artifact.uri, 'eval')
+        train_stats = tfx_util.read_stats_proto(stats_artifact.uri,
+                                                'Split-train')
+        eval_stats = tfx_util.read_stats_proto(stats_artifact.uri, 'Split-eval')
         graphics.annotate_dataset_feature_statistics_plots(
             model_card, [train_stats, eval_stats])
     return model_card

model_card_toolkit/template/html/default_template.html.jinja

@@ -41,6 +41,7 @@
 {% endmacro %}
 {% macro render_graphics(graphics) %}
   <div class="img-container">
+  {% if graphics.description %}<p>{{ graphics.description }}</p>{% endif %}
   {% for graph in graphics %}
     <div class="img-item">
       <img src='data:image/jpeg;base64,{{ graph.image }}' alt='{{ graph.name }}' />
@@ -131,7 +132,7 @@
         <h2>Model Details</h2>
           {% if model_details.overview %}<h3>Overview</h3>
               {{  model_details.overview }}{% endif %}
-          {% if model_details.version %}<h3>Version</h3>
+          {% if model_details.version and model_details.version.name %}<h3>Version</h3>
               {{ render_if_exist('name', model_details.version.name) }}
               {{ render_if_exist('date', model_details.version.date) }}
               {{ render_if_exist('diff', model_details.version.diff) }}