README.md

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/Mistral-7B-Instruct-v0.3-4bit") response = generate(model, tokenizer, prompt="hello", verbose=True)