apple-ml - Apple Silicon ML code patterns
Write PyTorch and MLX code for Apple Silicon M-series hardware without CUDA
Tags
Updated: 2026-03-25Capabilities
Typical Inputs
Typical Outputs
What this skill does
- select device
- set memory limit
- monitor memory
- synchronize device
- set fallback
- save checkpoint
- tune batch size
- schedule LR
- get memory
- set cache limit
- clear cache
- eval
- clip gradient
Inputs
- PyTorch code
- MLX code
- system memory
- training hyperparameters
- model weights
Outputs
- memory status
- device selection
- checkpoint file
- evaluation result
Requirements
- Apple Silicon M-series machine
- PyTorch
- MLX