graelo/Devstral-Small-2507-4bits

This model graelo/Devstral-Small-2507-4bits was converted to MLX format from mistralai/Devstral-Small-2507 using mlx-lm version 0.26.2.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("graelo/Devstral-Small-2507-4bits")

prompt = "hello"

if tokenizer.chat_template is not None:
    messages = [{"role": "user", "content": prompt}]
    prompt = tokenizer.apply_chat_template(
        messages, add_generation_prompt=True
    )

response = generate(model, tokenizer, prompt=prompt, verbose=True)

Important: use the feat/mistral branch from https://github.com/graelo/mlx-lm for direct support of Tekken in MLX-LM.