How To Export a TensorFlow Model#

Introduction#

In this guide you will learn how to export a TensorFlow model and register it in the Model Registry.

Code#

Step 1: Connect to Hopsworks#

import hopsworks

project = hopsworks.login()

# get Hopsworks Model Registry handle
mr = project.get_model_registry()

Step 2: Train#

Define your TensorFlow model and run the training loop.

# Define a model
model = tf.keras.Sequential()

# Add layers
model.add(..)

# Compile the model.
model.compile(..)

# Train the model
model.fit(..)

Step 3: Export to local path#

Export the TensorFlow model to a directory on the local filesystem.

model_dir = "./model"

tf.saved_model.save(model, model_dir)

Step 4: Register model in registry#

Use the ModelRegistry.tensorflow.create_model(..) function to register a model as a TensorFlow model. Define a name, and attach optional metrics for your model, then invoke the save() function with the parameter being the path to the local directory where the model was exported to.

# Model evaluation metrics
metrics = {'accuracy': 0.92}

tf_model = mr.tensorflow.create_model("tf_model", metrics=metrics)

tf_model.save(model_dir)

Conclusion#

In this guide you learned how to export a TensorFlow model to the Model Registry. You can also try attaching an Input Example and a Model Schema to your model to document the shape and type of the data the model was trained on.