Kurtis-E1.1-Qwen2.5-3B-Instruct-mlx-8Bit

linroger023/Kurtis-E1.1-Qwen2.5-3B-Instruct-mlx-8Bit

The Model linroger023/Kurtis-E1.1-Qwen2.5-3B-Instruct-mlx-8Bit was converted to MLX format from ethicalabs/Kurtis-E1.1-Qwen2.5-3B-Instruct using mlx-lm version 0.22.3.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("linroger023/Kurtis-E1.1-Qwen2.5-3B-Instruct-mlx-8Bit")

prompt="hello"

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

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