This package provides MATLAB® interfaces to MLflow™ Tracking and MLflow Model Logging. The package provides three main interfaces:
- Fluent interface to MLflow Tracking - this enables MATLAB to interact with MLflow Tracking using an interface which is very similar to the well known MLflow's Python fluent interface.
- MATLAB MLflow model flavor - this enables MATLAB code to be logged as MLflow models. These models can then later be loaded into MATLAB again and/or in some cases these models are then also useable outside of MATLAB.
- RESTful interface to MLflow Tracking - this enables MATLAB to interact with MLflow Tracking through its REST API. This interface is more verbose than the fluent interface but does not require Python®.
MathWorks Products (http://www.mathworks.com)
- MATLAB R2022b or later
- For selected features of the MATLAB MLflow model flavor:
- MATLAB Compiler SDK™, or
- Deep Learning Toolbox™, with
- Deep Learning Toolbox Converter for ONNX™ Model Format
- MLflow (compatible) Tracking servers and model registries, for example (but not limited to):
- MLflow Tracking Server
- Databricks™
- Microsoft Azure® Machine Learning
- For the fluent interface and MATLAB flavor
- Python version 3.10 or newer (also see Versions of Python Compatible with MATLAB Products by Release )
mlflowPython package version 2.* or 3.*
To use the package first clone the repository:
$ git clone https://github.com/mathworks-ref-arch/matlab-mlflow.gitThen consult Installation (fluent interface and MATLAB flavor) and/or Installation (RESTful interface) for further installation instructions.
Please see the documentation for more information.
The license for the MATLAB Interface for MLflow is available in the LICENSE.md file in this repository.
Provide suggestions for additional features or capabilities using the following link: https://www.mathworks.com/products/reference-architectures/request-new-reference-architectures.html
Please submit a GitHub issue.