A suite of tools that users, both novice and advanced, can use to optimize machine learning models for deployment and execution.

TensorFlow Model Optimization Toolkit

The TensorFlow Model Optimization Toolkit is a suite of tools that users, both novice and advanced, can use to optimize machine learning models for deployment and execution.

For an overview of this project and individual tools, the optimization gains, and our roadmap refer to tensorflow.org/model_optimization. The website also provides various tutorials and API docs.

The toolkit provides stable Python APIs.

Installation

Stable Builds

To install the latest version, run the following:

# Installing with the `--upgrade` flag ensures you'll get the latest version.
pip install --user --upgrade tensorflow-model-optimization

For release details, see our release notes.

For the required version of TensorFlow and other compatibility information, see the API Compatibility Matrix section of the Overview page for the technique you intend to use. For instance, for pruning, the Overview page is here.

Since TensorFlow is not included as a dependency of the TensorFlow Model Optimization package (in setup.py), you must explicitly install the TensorFlow package (tf-nightly or tf-nightly-gpu). This allows us to maintain one package instead of separate packages for CPU and GPU-enabled TensorFlow.

Installing from Source

You can also install from source. This requires the Bazel build system.

# To install dependencies on Ubuntu:
# sudo apt-get install bazel git python-pip
# For other platforms, see Bazel docs above.
git clone https://github.com/tensorflow/model-optimization.git
cd model_optimization
bazel build --copt=-O3 --copt=-march=native :pip_pkg
PKGDIR=$(mktemp -d)
./bazel-bin/pip_pkg $PKGDIR
pip install --user --upgrade $PKGDIR/*.whl

Contribution guidelines

If you want to contribute to TensorFlow Model Optimization, be sure to review the contribution guidelines. This project adheres to TensorFlow’s code of conduct. By participating, you are expected to uphold this code.

We use GitHub issues for tracking requests and bugs.

Community

As part of TensorFlow, we’re committed to fostering an open and welcoming environment.

• TensorFlow Blog: Stay up to date on content from the TensorFlow team and best articles from the community.