Description
Fine-tuning tutorials assume NVIDIA GPUs, leaving Mac users renting cloud servers. mlx-tune lets you fine-tune models natively on Apple Silicon Macs with MLX and an Unsloth-compatible API.
It supports SFT, DPO and GRPO plus vision, TTS, STT, embedding and OCR fine-tuning.
Native MLX:Train on Mac chips.
Methods:SFT, DPO and GRPO.
Multimodal:Vision, speech and OCR.
Unsloth-compatible:Familiar API.
It supports SFT, DPO and GRPO plus vision, TTS, STT, embedding and OCR fine-tuning.
Features
Native MLX:Train on Mac chips.
Methods:SFT, DPO and GRPO.
Multimodal:Vision, speech and OCR.
Unsloth-compatible:Familiar API.
