Skip to content
mlx.app

Unsloth

The fine-tuning library everyone quotes for faster training on less memory.

When to use it

Use this when you're training on CUDA hardware and plan to export the adapter to run on your Mac later — it isn't Apple-silicon native itself.

Alternatives in Fine-tuning

MLX LM LoRA

The same mlx-lm package doubles as Apple's official fine-tuning path.

Axolotl

The YAML-config trainer that treats every method as a config file.

PEFT

Hugging Face's parameter-efficient fine-tuning library, the one everything wraps.