Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
A native Mac App for LLM fine-tuning on Apple Silicon — fully on-device, fully open source.
| Date | Stars |
|---|---|
| 2026-07-24 | 250 |
| 2026-07-25 | 250 |
| 2026-07-28 | 250 |
| 2026-07-30 | 250 |
| 2026-07-31 | 251 |
| 2026-08-08 | 254 |
| 2026-08-16 | 256 |
| 2026-08-18 | 257 |
| 2026-08-26 | 258 |
| 2026-08-31 | 259 |
| 2026-09-02 | 260 |
| 2026-09-04 | 261 |
| 2026-09-06 | 262 |
| 2026-09-09 | 263 |
| 2026-09-11 | 264 |
| 2026-09-13 | 265 |
| 2026-09-17 | 267 |
| 2026-09-20 | 267 |
Today
— stars today
This week
+2 stars this week
This month
+10 stars this month
Momentum
37.0
growth rate 0.76%/day
<p align="center">
<img src="Sources/Media/logo_ultra-wide.png" alt="MLX LoRA Studio" width="100%">
</p>
<p align="center">
<strong>A native Mac App for LLM fine-tuning on Apple Silicon — fully on-device, fully open source.</strong>
</p>
<p align="center">
<em>
There is a quiet kind of power in running a model on the machine that sits in front of you.<br>
MLX LoRA Studio puts fine-tuning on your Mac — local, native, and visible end to end.<br>
Pick a model, choose an algorithm, watch the loss fall. The cloud is optional, the code is optional, the mystery is not.
</em>
</p>
<p align="center">
<img src="https://img.shields.io/badge/IMPORTANT-FIRST%20LAUNCH%20STEP-red?style=for-the-badge" alt="Important first launch step">
</p>
<div style="border: 4px solid #d1242f; background: #fff1f1; color: #8a1118; padding: 18px 22px; margin: 24px 0; font-size: 1.15rem;">
<strong>Important macOS first-launch step after installing:</strong><br><br>
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 <strong>not</strong> mean the app is malware. It is Apple's quarantine flag on a freshly downloaded app bundle.<br><br>
After dragging <strong>MLX LoRA Studio.app</strong> into <strong>/Applications</strong>, and before opening it for the first time, open Terminal and run:
<pre><code>sudo xattr -dr com.apple.quarantine "/Applications/MLX LoRA Studio.app"</code></pre>
Then open the app normally from <strong>/Applications</strong>. You only need to do this once per installed copy.
</div>
---
## Table of contents
- [Release v2.0.0](#release-v200)
- [Why MLX LoRA Studio?](#why-mlx-lora-studio)
- [What it is](#what-it-is)
- [What it isn't](#what-it-isnt)
- [Features at a glance](#features-at-a-glance)
- [Screenshots](#screenshots)
- [Installation](#installation)
- [Quick start (60 seconds)](#quick-start-60-seconds)
- [A guided tour of the app](#a-guided-tour-of-the-app)
- [Train](#train)
- [Live Metrics](#live-metrics)
- [Upload to HF](#upload-to-hf)
- [Algorithm Guide](#algorithm-guide)
- [Runs](#runs)
- [Settings & Onboarding](#settings--onboarding)
- [The training pipeline under the hood](#the-training-pipeline-under-the-hood)
- [Supported training methods](#supported-training-methods)
- [Configuration reference](#configuration-reference)
- [Memory & hardware expectations](#memory--hardware-expectations)
- [Building from source](#building-from-source)
- [Project layout](#project-layout)
- [Contributing](#contributing)
- [Star History](#star-history)
- [License](#license)
- [Acknowledgments](#acknowledgments)
---
## Release v2.0.0
**MLX LoRA Studio v2.0.0 updates the app for `mlx-lm-lora` 3.1.2.**
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` 3.1.2 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](https://goekdeniz-guelmez.github.io/MLX-LoRA-Studio/) 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 ResourcExcerpt of 24,976 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:d7bd1ebfdb000af9, topic:rlhf, topic:sft, desc:fine-tuning
matched fp:d7bd1ebfdb000af9, topic:reinforcement-learning, topic:rl