All Tools
M
LLMFreeOpen Source
MLX-LM
Run and fine-tune LLMs on Apple silicon with MLX
MIT
ABOUT
Running open LLMs on a Mac usually means CUDA stacks that do not exist on Apple silicon. mlx-lm generates, chats, quantizes, and LoRA-tunes Hugging Face models with MLX so Mac users can serve and adapt models locally without a GPU server.
INSTALL
pip install mlx-lmINTEGRATION GUIDE
1. Chat with a 4-bit Llama or Mistral model on Apple silicon from the CLI
2. Quantize a Hugging Face checkpoint and upload it to the MLX Community hub
3. LoRA fine-tune an MLX model on a Mac without CUDA
4. Cache a long prompt and reuse it across generate calls for multi-turn chat
TAGS
pythonmlxapple-siliconllmquantizationlorahuggingface