mlx-community/deepseek-coder-v2-lite-instruct-4bit-awq

This model mlx-community/deepseek-coder-v2-lite-instruct-4bit-awq was converted to MLX format from deepseek-ai/DeepSeek-Coder-V2-Lite-Instruct using mlx-lm version 0.25.2.

AWQ Parameters: --bits 4 --group-size 64 --embed-bits 4 --embed-group-size 32 --num-samples 128 --sequence-length 1024 --n-grid 50 with a customized calibration dataset

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/deepseek-coder-v2-lite-instruct-4bit-awq")

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)