DNNClassifier model to TensorFlow Serving model

Question

I am a newbie in ML and TF and I am trying to host primitive TensorFlow model on GCP using TensorFlow Serving. For do that I need to convert DNNClassifier model to TensorFlow Serving model. According to Get Started guide I need to use SavedModelBuilder method but I can't figure out how to define input/outputs in the case with Iris Flower example.

Could anybody post an example code for this case?

Full code:

(train_x, train_y), (test_x, test_y) = iris_data.load_data()

# Feature columns describe how to use the input.
my_feature_columns = []
for key in train_x.keys():
    my_feature_columns.append(tf.feature_column.numeric_column(key=key))

# Build 2 hidden layer DNN with 10, 10 units respectively.
classifier = tf.estimator.DNNClassifier(
    feature_columns=my_feature_columns,
    # Two hidden layers of 10 nodes each.
    hidden_units=[10, 10],
    # The model must choose between 3 classes.
    n_classes=3)

# Train the Model.
classifier.train(
    input_fn=lambda:iris_data.train_input_fn(train_x, train_y,
                                             args.batch_size),
    steps=args.train_steps)

# Evaluate the model.
eval_result = classifier.evaluate(
    input_fn=lambda:iris_data.eval_input_fn(test_x, test_y,
                                            args.batch_size))

print('\nTest set accuracy: {accuracy:0.3f}\n'.format(**eval_result))

# Generate predictions from the model
expected = ['Setosa', 'Versicolor', 'Virginica']
predict_x = {
    'SepalLength': [5.1, 5.9, 6.9],
    'SepalWidth': [3.3, 3.0, 3.1],
    'PetalLength': [1.7, 4.2, 5.4],
    'PetalWidth': [0.5, 1.5, 2.1],
}

predictions = classifier.predict(
    input_fn=lambda:iris_data.eval_input_fn(predict_x,
                                            labels=None,
                                            batch_size=args.batch_size))

for pred_dict, expec in zip(predictions, expected):
    template = ('\nPrediction is "{}" ({:.1f}%), expected "{}"')

    class_id = pred_dict['class_ids'][0]
    probability = pred_dict['probabilities'][class_id]

    print(template.format(iris_data.SPECIES[class_id],
                          100 * probability, expec))

Answer 1

Just right after training and evaluating your model, you are able to save the model.

(train_x, train_y), (test_x, test_y) = iris_data.load_data()

# Feature columns describe how to use the input.
my_feature_columns = []
for key in train_x.keys():
    my_feature_columns.append(tf.feature_column.numeric_column(key=key))

# Build 2 hidden layer DNN with 10, 10 units respectively.
classifier = tf.estimator.DNNClassifier(
    feature_columns=my_feature_columns,
    # Two hidden layers of 10 nodes each.
    hidden_units=[10, 10],
    # The model must choose between 3 classes.
    n_classes=3)

# Train the Model.
classifier.train(
    input_fn=lambda:iris_data.train_input_fn(train_x, train_y,
                                             args.batch_size),
    steps=args.train_steps)

# Evaluate the model.
eval_result = classifier.evaluate(
    input_fn=lambda:iris_data.eval_input_fn(test_x, test_y,
                                            args.batch_size))

export_path = 'Your Desired new Path '
builder = tf.saved_model.builder.SavedModelBuilder(export_path)
sess = tf.InteractiveSession()
builder.add_meta_graph_and_variables(
  sess, [tf.saved_model.tag_constants.SERVING]
builder.save()

Depending on your application, you can also addsignature_def_map to builder.add_meta_graph_and_variables() function.

Please note that for the classifier the input is feature_columns and the output is one of the three classes. For Builder, the input is 'tf session,tag_constants.SERVINGandsignature_def_map` and the output is 'Desired_Directory/saved_model.pb'

Answer 2

Just change the arythmic pattern to a tensor style might have to cross incorporate styles then use a format equalizer for adjustments.