Skip to main content
Version: 0.7.0

Submarine Python SDK

Submarine Python SDK can runs on any machine and it will talk to Submarine Server via REST API. So you can install Submarine Python SDK on your laptop, a gateway machine, your favorite IDE (like PyCharm/Jupyter, etc.).

Furthermore, Submarine supports an extensible package of CTR models based on TensorFlow and PyTorch along with lots of core components layers that can be used to easily build custom models. You can train any model with model.train() and model.predict().

Prepare Python Environment to run Submarine SDK​

Submarine SDK requires Python3.7+. It's better to use a new Python environment created by Anoconda or Python virtualenv to try this to avoid trouble to existing Python environment. A sample Python virtual env can be setup like this:

tar xf virtualenv-16.0.0.tar.gz

# Make sure to install using Python 3
python3 virtualenv-16.0.0/ venv
. venv/bin/activate

Install Submarine SDK​

Starting from 0.4.0, Submarine provides Python SDK. Please change it to a proper version needed.

More detail:

# Install latest stable version
pip install apache-submarine
# Install specific version
pip install apache-submarine==<REPLACE_VERSION>

Install SDK from source code​

Please first clone code from github or go to to download released source code.

git clone
# (optional) chackout specific branch or release
git checkout <correct release tag/branch>
cd submarine/submarine-sdk/pysubmarine
pip install .

Manage Submarine Experiment​

Assuming you've installed submarine on K8s and forward the traefik service to localhost, now you can open a Python shell, Jupyter notebook or any tools with Submarine SDK installed.

Follow SDK experiment example to run an experiment.

Training a DeepFM model​

The Submarine also supports users to train an easy-to-use CTR model with a few lines of code and a configuration file, so they don’t need to reimplement the model by themself. In addition, they can train the model on both local on distributed systems, such as Hadoop or Kubernetes.

Follow SDK DeepFM example to try the model.