# MLX-LM Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/mlx-lm
> Markdown URL: https://aitinkerers.org/technologies/mlx-lm.md
> Technology record last updated: 2026-02-26T11:12:56Z
> Generated: 2026-09-22T22:38:23Z

The Python package for efficient text generation and fine-tuning of Large Language Models (LLMs) directly on Apple silicon via the MLX framework.

MLX-LM is a high-performance Python package engineered for text generation and fine-tuning of Large Language Models (LLMs) on Apple silicon, leveraging the core MLX array framework. It provides seamless integration with the Hugging Face Hub, allowing users to easily access and run thousands of LLMs with a single command. Key features include native support for 4-bit quantization to reduce model memory footprint and efficient low-rank or full model fine-tuning. This package enables developers to maximize the unified memory architecture of Apple silicon for faster, on-device machine learning workflows.

- Official technology site: https://github.com/ml-explore/mlx-lm
- Public AI Tinkerers demos and talks: 5
- Result page: 1 of 1

## Recent Public Talks and Demos

### [AI for Smart Home control](https://ottawa.aitinkerers.org/talks/rsvp_TCYimty3wlo)

I have built an MVP of a complete home control solution. It supports control of devices in the home like thermostats, lights, blinds... using the Matter protocol and a local AI engine. The local AI is based on Gemma4 E4B for the embedded Jetson Orin Nano solution and Qween 3.5 9B for the Mac M4 MIni based solution. The solution supports a voice pipeline with Whisper STT and Kokoro TTS and allows the user to converse with the AI agent to setup automations, control devices... The MAC solutions uses mlx-lm and is optimized with prompt caching to reduce response times. The voice pipeline supports multi-turn conversations. The solution also includes an iPhone and Android based app to control the home and converse with the LLM.

- Event context: AI Tinkerers Ottawa Meetup - June 20th, 2026 — 2026-06-20 — Ottawa
- Public talk page: https://ottawa.aitinkerers.org/talks/rsvp_TCYimty3wlo

### [Six steps to make your agent smarter using small local models.](https://columbus.aitinkerers.org/talks/rsvp_3wl4rNO70ks)

Yeah, I built out a six-step process on a Mac Mini by which all agent turns are recorded, judged, scored, and then used as training material to take a qwin 2.5B model and make it as good as OpenAI 5.4.

- Event context: AI Tinkerers - Columbus May Meetup — 2026-05-04 — Columbus
- Public talk page: https://columbus.aitinkerers.org/talks/rsvp_3wl4rNO70ks

### [Eric Chat: Run AI models locally, securly and offline on Macs](https://ottawa.aitinkerers.org/talks/rsvp_3gOvbv-b9DU)

Eric Chat is a Python package that lets users run models up to 120 billion parameters offline on Macs with Apple Silicon. It provides an easy-to-use graphical user interface.

- Event context: AI Tinkerers Ottawa Meetup — April 25th, 2026 — 2026-04-25 — Ottawa
- Public talk page: https://ottawa.aitinkerers.org/talks/rsvp_3gOvbv-b9DU

### [Fine-Tuning Models for Content Moderation with Apple’s MLX Framework](https://orange-county.aitinkerers.org/talks/rsvp_S0EeHEPZUv4)

This talk is a hands-on walkthrough of how to fine-tune large language models—and multi-modal models—locally on Apple-Silicon Macs using Apple’s open-source MLX framework. We’ll: 1. Give a quick MLX overview and why it’s optimized for the M-series GPU/ANE. 2. Explain LoRA / QLoRA and why adapter-based fine-tuning is memory-efficient. 3. Dissect the vision-encoder → adapter → LLM pipeline (SigLIP + Phi-1.5). 4. Show dataset prep in JSONL, CLI commands (mlx_lm.lora, mlx_lm.fuse) and YAML options. 5. Live-interpret training logs, validation curves, and evaluation metrics. 6. End with best-practice checklists for scaling from small experiments to full production runs. Attendees will leave able to replicate the full workflow—dataset → training → evaluation → deployment—entirely on their MacBooks.

- Event context: AI Tinkerers - Orange County Meetup- Wednesday June 4th 2025 — 2025-06-05 — Orange County
- Public talk page: https://orange-county.aitinkerers.org/talks/rsvp_S0EeHEPZUv4

### [cupertino.ink](https://singapore.aitinkerers.org/talks/rsvp_JS1WBDkSB1c)

A pure local server that executes various tools from MCP providers as well as local MCPs, running using a custom MLX engine I made myself(For people with Macs)

- Event context: AI Tinkerers Singapore: 6th Meetup - April 25th, 2025 — 2025-04-25 — Singapore
- Public talk page: https://singapore.aitinkerers.org/talks/rsvp_JS1WBDkSB1c

## Related Technologies

- [Python](https://aitinkerers.org/technologies/python) ([Markdown](https://aitinkerers.org/technologies/python.md)) — 662 public demos
- [A2A](https://aitinkerers.org/technologies/a2a) ([Markdown](https://aitinkerers.org/technologies/a2a.md)) — 7 public demos
- [Apple Silicon](https://aitinkerers.org/technologies/apple-silicon) ([Markdown](https://aitinkerers.org/technologies/apple-silicon.md)) — 2 public demos
- [Eric Transformer](https://aitinkerers.org/technologies/eric-transformer) ([Markdown](https://aitinkerers.org/technologies/eric-transformer.md)) — 1 public demo
- [gemma4:e4b](https://aitinkerers.org/technologies/gemma4-e4b) ([Markdown](https://aitinkerers.org/technologies/gemma4-e4b.md)) — 1 public demo
- [Kokoro TTS](https://aitinkerers.org/technologies/kokoro-tts) ([Markdown](https://aitinkerers.org/technologies/kokoro-tts.md)) — 2 public demos
- [LoRA](https://aitinkerers.org/technologies/lora) ([Markdown](https://aitinkerers.org/technologies/lora.md)) — 17 public demos
- [Mac Mini](https://aitinkerers.org/technologies/mac-mini) ([Markdown](https://aitinkerers.org/technologies/mac-mini.md)) — 3 public demos
- [MLX](https://aitinkerers.org/technologies/mlx) ([Markdown](https://aitinkerers.org/technologies/mlx.md)) — 7 public demos
- [OpenAI](https://aitinkerers.org/technologies/openai) ([Markdown](https://aitinkerers.org/technologies/openai.md)) — 112 public demos
- [QLoRA](https://aitinkerers.org/technologies/qlora) ([Markdown](https://aitinkerers.org/technologies/qlora.md)) — 4 public demos
- [Qween 3](https://aitinkerers.org/technologies/qween-3) ([Markdown](https://aitinkerers.org/technologies/qween-3.md)) — 1 public demo
- [Qwen](https://aitinkerers.org/technologies/qwen) ([Markdown](https://aitinkerers.org/technologies/qwen.md)) — 18 public demos
- [Qwin 2](https://aitinkerers.org/technologies/qwin-2) ([Markdown](https://aitinkerers.org/technologies/qwin-2.md)) — 1 public demo
- [Toga](https://aitinkerers.org/technologies/toga) ([Markdown](https://aitinkerers.org/technologies/toga.md)) — 1 public demo
- [TurboAPI](https://aitinkerers.org/technologies/turboapi) ([Markdown](https://aitinkerers.org/technologies/turboapi.md)) — 1 public demo
- [Whisper STT](https://aitinkerers.org/technologies/whisper-stt) ([Markdown](https://aitinkerers.org/technologies/whisper-stt.md)) — 1 public demo