Agent Skills: mlx-fine-tuning

Specialized skill for MLX-based LLM fine-tuning on Apple Silicon. This skill should be used when working with LoRA fine-tuning, model conversion from HuggingFace to MLX format, adapter management, hyperparameter tuning, memory optimization, and benchmarking on M1/M2/M3/M4 chips.