Valkyrie-49B-v2.1-mlx-4Bit

ailexleon/Valkyrie-49B-v2.1-mlx-4Bit

Note: To load the model using the latest versions, you must set trust_remote_code=True in python, use --trust-remote-code with mlx_lm or enable "Trust Remote Code" in the model settings in oMLX. There is currently no way to do this with LMStudio.

The Model ailexleon/Valkyrie-49B-v2.1-mlx-4Bit was converted to MLX format from TheDrummer/Valkyrie-49B-v2.1 using mlx-lm version 0.29.1.

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("ailexleon/Valkyrie-49B-v2.1-mlx-4Bit")

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)