Core ML Tools API Overview

Core ML Tools is the coremltools Python package for macOS (10.13+) and Linux. It includes the Unified Conversion API for converting deep learning models and neural networks to Core ML.

For example, you can use the Unified Conversion API to convert TensorFlow and PyTorch source model frameworks to Core ML. For the conversion parameters, see the convert() method.

Note

This section is about converting neural network models using the Unified Conversion API. For converting other classic models, see LibSVM, Scikit-learn, and XGBoost in the “Other Converters” section.

For instructions and examples, see the following pages:

• Converting Deep Learning Models

• ML Programs

• Converting from PyTorch

• Converting from TensorFlow

• Examples

For common scenarios using conversion options, see the following pages:

• Model Input and Output Types

• Image Input and Output

• Classifiers

• Flexible Input Shapes

• Composite Operators

• Custom Operators

• Graph Passes