Load and run Core ML models on-device in Swift iOS apps.
npx skills add https://github.com/Borisserz/FoodTracker --skill coreml-borisserz
Please help me install this Agent Skill. Skill: coreml Source: https://github.com/Borisserz/FoodTracker/tree/main/.agents/skills/coreml Command: npx skills add https://github.com/Borisserz/FoodTracker --skill coreml-borisserz
Developers need a clear, end-to-end guide to integrating Core ML models on-device in Swift-based iOS apps, including loading, configuration, prediction, and deployment considerations.
Load a Core ML model into an iOS project and run a first prediction to verify the integration.
High-intent search queries and answers about installing and using this skill.
To run Core ML on-device predictions in Swift, load .mlmodel, .mlpackage, or .mlmodelc formats using auto-generated classes or MLFeatureProvider. You can then configure compute units via MLModelConfiguration before executing inference directly within your iOS application.
MLComputePlan analyzes Core ML model execution to profile performance by evaluating the configured compute units. It helps developers assess deployment strategies and understand inference costs before running predictions on the device's CPU, GPU, or Neural Engine.
Yes, you can build multi-model pipelines in iOS 14+ and Swift 6.x by integrating the Vision framework with Core ML. This combination enables efficient on-device inference workflows and image processing tasks using your deployed models.
The best way to configure compute units for an .mlpackage model is using MLModelConfiguration. This allows you to selectively target the CPU, GPU, or Neural Engine to optimize on-device inference performance based on your specific deployment requirements.
Yes, Core ML supports loading .mlmodel, .mlpackage, and compiled .mlmodelc formats for on-device inference. You can execute predictions using either auto-generated Swift classes or the MLFeatureProvider interface.
MLFeatureProvider is needed for Core ML predictions when building dynamic multi-model pipelines or handling inputs not covered by auto-generated classes. It provides a flexible interface to manage feature inputs and outputs for on-device inference execution.
harshav167Load Core ML models and run predictions in Swift iOS apps.
ckorhonenOptimize CoreML models with quantization, palettization, and pruning.
femitzLoad and run Core ML models on-device with Swift APIs.
LaughingJackalopeIntegrates Apple Foundation Models, CoreML, and MLX Swift for on-device AI in macOS 26 apps.
