Welcome to tensorflow.m’s documentation!

tensorflow.m is a free and open-source MATLAB/Octave package, interfacing TensorFlow’s C API without requiring any additional tools.

Looking for an easy way to get started? Check out the quick start guide.

This allows you to use the functionality of TensorFlow’s core from within MATLAB/Octave via a MEX function and wrapper classes. Since the purpose of this package is to interface libraries provided by TensorFlow (rather than duplicating code in another language), anything not exposed to the C API is considered beyond the scope. As an example, modules like tf.estimator or tf.keras in Python are not accessible in tensorflow.m.

Compatibility

tensorflow.m is written to work for recent versions of MATLAB (tested on R2019b; Windows, macOS and Linux) and Octave. Unit tests are run against Octave 4.4.1 and 5.1.0 on a Debian x86-64 image.

License

tensorflow.m is licensed under the Apache License 2.0.

Note: This software is in active development, we cannot make any promises regarding the stability of the API.