Machine learning Interview

from mlx_lm import load, generate

model, tokenizer = load("mlx-community/DeepSeek-R1-0528-4bit")

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)

👉 huggingface.co/mlx-community/DeepSeek-R1-0528-4bit

#DeepSeek

www.tg-me.com/it/Machine learning Interview/com.machinelearning_interview/1820

3.1K viewsMay 29 at 13:50

tg-me.com/machinelearning_interview/1820

Create: 2025-05-29
Last Update: 2025-06-15 01:16:05

🔥 DeepSeek R1 (0528) — компактная MLX-версия!

А вот и 4-битная квантованная версия DeepSeek R1 специально под MLX.

Модель запускается локально на Mac с Apple Silicon без танцев с бубном и GPU.

Установка:

pip install mlx-lm

from mlx_lm import load, generate model, tokenizer = load("mlx-community/DeepSeek-R1-0528-4bit") 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)

👉 huggingface.co/mlx-community/DeepSeek-R1-0528-4bit

#DeepSeek

BY Machine learning Interview

Share with your friend now:
tg-me.com/machinelearning_interview/1820

Machine learning Interview Telegram | DID YOU KNOW?