# MLX LoRA Studio

A native macOS application for fine-tuning large language models on Apple Silicon, letting a user pick a model and a training algorithm and watch the training loss fall entirely on-device, with no cloud step required.

MLX LoRA Studio is a SwiftUI and AppKit application, built on the mlx-lm-lora library, that runs a full local fine-tuning workflow for MLX-compatible language models on a Mac. It supports ten training algorithms, including SFT, DPO, FTPO, CPO, ORPO, GRPO, Online DPO, XPO, RLHF Reinforce and PPO, several training modes such as LoRA, DoRA and QLoRA at multiple bit widths, and dynamic SFT objectives including memory-bounded chunked negative-log-likelihood training for constrained hardware.

A guided tour covers training, live training metrics, uploading a result to the Hugging Face Hub, an algorithm guide, a history of past runs, and settings and onboarding. Because the first open-source release is distributed outside the Mac App Store and is not notarized, macOS shows a quarantine warning on first launch that a documented terminal command clears; this is unrelated to the app's own functionality.

Website: https://mlx-lora-studio.netlify.app