Deepseek R1 Distill Llama 70B Abliterated Mlx 4Bit

🚀 cs2764/DeepSeek-R1-Distill-Llama-70B-abliterated-mlx-4Bit

This project focuses on converting the model into the MLX format, enabling more efficient and flexible use in machine learning applications. It provides a convenient way to utilize the model's capabilities with the MLX library.

🚀 Quick Start

The Model cs2764/DeepSeek-R1-Distill-Llama-70B-abliterated-mlx-4Bit was converted to MLX format from huihui-ai/DeepSeek-R1-Distill-Llama-70B-abliterated using mlx-lm version 0.22.3.

📦 Installation

pip install mlx-lm

💻 Usage Examples

Basic Usage

from mlx_lm import load, generate

model, tokenizer = load("cs2764/DeepSeek-R1-Distill-Llama-70B-abliterated-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)

📚 Documentation