# M-Courtyard: GUI app for fine-tuning mlx-community models on Mac (open source)

**URL:** <https://discuss.huggingface.co/t/m-courtyard-gui-app-for-fine-tuning-mlx-community-models-on-mac-open-source/173454>\
**Category:** Beginners\
**Created:** [February 13, 2026, 12:32pm UTC](https://discuss.huggingface.co/t/m-courtyard-gui-app-for-fine-tuning-mlx-community-models-on-mac-open-source/173454 "2026-02-13T12:32:49Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![tuwenbo](https://avatars.discourse-cdn.com/v4/letter/t/e99b99/32.png) [@tuwenbo](https://discuss.huggingface.co/u/tuwenbo)\
**Post date:** [February 13, 2026, 12:32pm UTC](https://discuss.huggingface.co/t/m-courtyard-gui-app-for-fine-tuning-mlx-community-models-on-mac-open-source/173454/1 "2026-02-13T12:32:49Z")

</div>

Hi everyone!

I built an open-source macOS app called M-Courtyard that provides a full GUI  
for fine-tuning LLMs using mlx-lm on Apple Silicon.

It integrates directly with Hugging Face’s mlx-community models — you can  
browse, download, and fine-tune models like Qwen 3, DeepSeek R1, Llama 3  
and more, all through a visual interface.

Workflow:

1. Import documents → Auto-generate training data
2. Select a model from mlx-community (auto-download)
3. Fine-tune with LoRA/DoRA — real-time loss visualization
4. Test with built-in chat
5. Export to Ollama

No CLI, no Python scripts needed.

GitHub: [GitHub - tuwenbo0120/m-courtyard: M-Courtyard: Local AI Model Fine-tuning Assistant for Apple Silicon. Zero-code, zero-cloud, privacy-first desktop app powered by Tauri + React + mlx-lm.](https://github.com/tuwenbo0120/m-courtyard)  
Requirements: macOS 14+, Apple Silicon, 16GB+ RAM recommended

Would love to hear your thoughts, especially from fellow MLX users!