GLM-4.5-Air-q3-hi-mlx

This quant was converted with page size 32.

It is a bit larger than the default q3 quant made with page size 64, and should perform slightly better.

I will also upload an xs quant made with page size 128, to see the difference.

Similar quant versions will be available for q4, q5, and q6

This model GLM-4.5-Air-q3-hi-mlx was converted to MLX format from zai-org/GLM-4.5-Air using mlx-lm version 0.26.1.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("GLM-4.5-Air-q3-hi-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)