drewbenson/DeepSeek-R1-Distill-Llama-8B-4bit-MLX

The Model drewbenson/DeepSeek-R1-Distill-Llama-8B-4bit-MLX was converted to MLX format from deepseek-ai/DeepSeek-R1-Distill-Llama-8B using mlx-lm version 0.21.5.

On M4 Max, this runs 40%-50% faster than the GGUF.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("drewbenson/DeepSeek-R1-Distill-Llama-8B-4bit-MLX")

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)