mlx-tune
Fine-tune LLMs on your Mac with Apple Silicon for various tasks including SFT, DPO, GRPO, Vision, TTS, STT, Embedding, and OCR.
GraphCanon updated 3w · GitHub synced 3w
Decision brief
mlx-tune targets Mac users with Apple Silicon for fine-tuning LLMs across SFT, RLHP, GRPO, vision, TTS, STT, embeddings, and OCR using tools compatible with the UnSloth API.
Good fit when
- You need to fine-tune large language models on a Mac with Apple Silicon hardware
- You aim to work specifically with tasks like SFT, RLHP, GRPO, vision-based, TTS, STT, embeddings generation, or OCR
Avoid when
- Your development environment is not based on macOS running on Apple Silicon
- The specific tasks you are targeting do not align with the capabilities of mlx-tune such as those exclusive to alternative platforms or tools
Observed Jul 14, 2026 · Source: enrich:decision_facts
Verify the decision
Maintenance and security
Full trust report- Maintenance
- Steady (36d since push)
- As of 3w
- Provenance
- Not a fork · Personal account
- As of 3w
- Security (OSV)
- 46 low (46 low)
- As of 1mo
Public GitHub metadata and optional OSV scans. Signals, not a guarantee. Trust methodology.
Install
pip install mlx-tuneSimilar tools
Same-category neighbours. No typed graph edges are catalogued for this tool yet.
Evidence and technical details
Sourced facts, taxonomy, compatibility claims, README excerpt, and machine-readable endpoints.
Overview
mlx-tune offers tools to fine-tune large language models using Apple Silicon. Supports a wide range of tasks such as supervised fine-tuning (SFT), reinforcement learning from human preferences (RLHP), generative pre-trained transformer (GPT) retraining on prompt optimization (GRPO), vision, text-to-speech (TTS), speech-to-text (STT), embeddings generation, and optical character recognition (OCR). Framework is compatible with the UnSloth API.
Capability facts
- Languages
- python
Source: github.language+pyproject.toml · Jul 30, 2026
Categories
Compatibility
Sourced claims from the README excerpt - not unsourced marketing copy.
Source: README excerpt (regex_v1, Jul 30, 2026)
```python from mlx_tune import FastLanguageModel, SFTTrainer, SFTConfigSource link
Tags
README
Quick Start
from mlx_tune import FastLanguageModel, SFTTrainer, SFTConfig
from datasets import load_dataset
---
# Load any HuggingFace model (1B model for quick start)
model, tokenizer = FastLanguageModel.from_pretrained(
model_name="mlx-community/Llama-3.2-1B-Instruct-4bit",
max_seq_length=2048,
load_in_4bit=True,
)
---
## Requirements
- **Hardware**: Apple Silicon Mac (M1/M2/M3/M4/M5)
- **OS**: macOS 13.0+
- **Memory**: 8GB+ unified RAM (16GB+ recommended)
- **Python**: 3.9+
---
## License
Apache 2.0 - See [LICENSE](LICENSE) file.
For agents
This page has a .md twin and JSON over the API.