Optimize CoreML models with quantization, palettization, and pruning.
npx skills add https://github.com/ckorhonen/claude-skills --skill coreml-optimizer
Please help me install this Agent Skill. Skill: coreml-optimizer Source: https://github.com/ckorhonen/claude-skills/tree/main/skills/coreml-optimizer Command: npx skills add https://github.com/ckorhonen/claude-skills --skill coreml-optimizer
CoreML models often require optimization to reduce file size and improve on-device latency, making on-device inference faster and cheaper.
Run the CoreML optimization workflow to compress weights and accelerate on-device inference.
High-intent search queries and answers about installing and using this skill.
Optimize CoreML models by applying quantization, palettization, and pruning to compress weights. Converting models to the ML program format reduces on-device file size and accelerates inference latency.
Target the Apple Neural Engine by configuring compute_units settings during ML program deployment. Pre-execution analysis with MLComputePlan and model introspection helps map operations to the Neural Engine, CPU, or GPU.
Profile and debug CoreML performance using MLComputePlan, Core ML Instruments, and model introspection. These tools analyze on-device execution, identify compute unit bottlenecks, and evaluate pre-execution performance bottlenecks.
Converting to the ML program format improves CoreML inference speed by enabling optimized compute_units configuration. This format supports advanced weight compression and efficient execution across the Neural Engine, CPU, and GPU.
Apply quantization or palettization to a neural network model when you need to reduce on-device size and lower latency while preserving accuracy. These compression techniques optimize weights for efficient on-device inference.
pradeepmouliConvert PyTorch models to CoreML and optimize for iOS/macOS deployment.
jperezdelrealLoad Core ML models and run predictions with Swift interfaces.
Roy-wonjiIntegrate Core ML models into iOS apps for on-device inference.
hung-phanPrototype and train neural networks using high-level Keras APIs.
