StackMap
Subscribe
Explore / MLX-LoRA-Studio
Goekdeniz-Guelmez

MLX-LoRA-Studio

Native Mac app for on-device LLM fine-tuning via mlx-lm-lora: pick a model, choose SFT/LoRA/DPO-family algorithms, watch loss fall live, push to Hugging Face. No cloud, no code.

250 24 Swift MITupdated 4 days ago
traininglocal
View on GitHubDispute this mapping →
Curator's take

The shortest path from base model to tuned adapter on a Mac: no Python env, no GPU rental — a native SwiftUI app driving mlx-lm-lora, with an algorithm menu unusually deep for a GUI (SFT through DPO-family preference methods, FTPO, Dynamic Fine-Tuning, memory-bounded Chunked NLL), live loss curves and direct HF upload. NOT for serious scale: your ceiling is unified memory on one machine — multi-GPU and large post-training runs belong in LLaMA-Factory/TRL territory. Young single-maintainer project (~250 stars) with a quarantine-xattr dance on first launch; Apple Silicon only, by design.

Mapped by ShipWithAI editors · links verified
README.md

MLX LoRA Studio

A native Mac App for LLM fine-tuning on Apple Silicon — fully on-device, fully open source.

There is a quiet kind of power in running a model on the machine that sits in front of you.
MLX LoRA Studio puts fine-tuning on your Mac — local, native, and visible end to end.
Pick a model, choose an algorithm, watch the loss fall. The cloud is optional, the code is optional, the mystery is not.

Important first launch step

Important macOS first-launch step after installing:

Because this first open-source release is distributed outside the Mac App Store and is not yet notarized, macOS may show a scary "damaged", "cannot be opened", or unidentified developer warning. This does not mean the app is malware. It is Apple's quarantine flag on a freshly downloaded app bundle.

After dragging MLX LoRA Studio.app into /Applications, and before opening it for the first time, open Terminal and run: 
sudo xattr -dr com.apple.quarantine "/Applications/MLX LoRA Studio.app"
Then open the app normally from /Applications. You only need to do this once per installed copy.

Table of contents

Release v2.0.0

MLX LoRA Studio v2.0.0 updates the app for mlx-lm-lora 3.0.0.

This release adds FTPO, Dynamic Fine-Tuning, and memory-bounded Chunked NLL while aligning the native training runner and documentation with the mlx-lm-lora 3.0.0 API. Synthetic dataset creation has been removed from the active app workflow, while historical synthetic run folders remain readable in the Runs archive.

If you need detialed and longer explanations for the algorythms used, then visit the wiki page.

What's included

  • Native macOS app for Apple Silicon built with SwiftUI and AppKit.
  • Fully local fine-tuning workflow for MLX-compatible language models.
  • 10 training algorithms: SFT, DPO, FTPO, CPO, ORPO, GRPO, Online DPO, XPO, RLHF Reinforce, and PPO.
  • Dynamic SFT objectives: NLL, memory-bounded Chunked NLL, and Dynamic Fine-Tuning.
  • Multiple training modes: LoRA, DoRA, QLoRA at 4/6/8-bit, full fine-tuning, and Quantization-Aware Training (QAT).
  • Live training observability with loss, learning rate, gradient norm, throughput, progress, logs, and recent-step charts.
  • Memory-aware run planning with live wired/active memory monitoring and Resourc

Continue your stack

What teams reach for next — and why each earns a place beside MLX-LoRA-Studio. Ranked by curator confidence.