Core ML Tools

iOS

Core ML Tools is a Python package designed to convert third-party models (PyTorch, TensorFlow) to the Core ML model package format, allowing integration into iOS apps. It also provides APIs for optimizing models to use less storage space, reduce power consumption, and reduce latency during interference.

Using the Unified Conversion API, Core ML Tools converts a third-party model to a more malleable representation in Model Intermediate Language (MIL) before translating to the Core ML format. This allows for adjustments like fine-tuning for specific hardware capabilities and applying quantization for model size reduction before deploying the final, optimized Core ML model to an iOS app.

Converting a third-party model to an MIL program, and then to a neural network or ML program

Core ML Tools also supports compression workflows for reducing the size of models. In Core ML Tools 7, there is support for post-training compression (for all models) and training time compression (for PyTorch models). The first approach is faster and doesn't require data, whereas the second offers better accuracy retention by leveraging data for fine-tuning.

For more information, check out Apple’s Core ML Tools guide.

Sign up for our newsletter 🛸

Never miss a post or product update