Instructions to use cudo528/Darwin-35B-A3B-Opus-mlx-8bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use cudo528/Darwin-35B-A3B-Opus-mlx-8bit with MLX:
# Make sure mlx-vlm is installed # pip install --upgrade mlx-vlm from mlx_vlm import load, generate from mlx_vlm.prompt_utils import apply_chat_template from mlx_vlm.utils import load_config # Load the model model, processor = load("cudo528/Darwin-35B-A3B-Opus-mlx-8bit") config = load_config("cudo528/Darwin-35B-A3B-Opus-mlx-8bit") # Prepare input image = ["http://images.cocodataset.org/val2017/000000039769.jpg"] prompt = "Describe this image." # Apply chat template formatted_prompt = apply_chat_template( processor, config, prompt, num_images=1 ) # Generate output output = generate(model, processor, formatted_prompt, image) print(output) - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
- Pi
How to use cudo528/Darwin-35B-A3B-Opus-mlx-8bit with Pi:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "cudo528/Darwin-35B-A3B-Opus-mlx-8bit"
Configure the model in Pi
# Install Pi: npm install -g @earendil-works/pi-coding-agent # Add to ~/.pi/agent/models.json: { "providers": { "mlx-lm": { "baseUrl": "http://localhost:8080/v1", "api": "openai-completions", "apiKey": "none", "models": [ { "id": "cudo528/Darwin-35B-A3B-Opus-mlx-8bit" } ] } } }Run Pi
# Start Pi in your project directory: pi
- Hermes Agent
How to use cudo528/Darwin-35B-A3B-Opus-mlx-8bit with Hermes Agent:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "cudo528/Darwin-35B-A3B-Opus-mlx-8bit"
Configure Hermes
# Install Hermes: curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash hermes setup # Point Hermes at the local server: hermes config set model.provider custom hermes config set model.base_url http://127.0.0.1:8080/v1 hermes config set model.default cudo528/Darwin-35B-A3B-Opus-mlx-8bit
Run Hermes
hermes
- Atomic Chat
- OpenClaw
How to use cudo528/Darwin-35B-A3B-Opus-mlx-8bit with OpenClaw:
Start the MLX server
# Install MLX LM: uv tool install mlx-lm # Start a local OpenAI-compatible server: mlx_lm.server --model "cudo528/Darwin-35B-A3B-Opus-mlx-8bit"
Configure OpenClaw
# Install OpenClaw: npm install -g openclaw@latest # Register the local server and set it as the default model: openclaw onboard --non-interactive --mode local \ --auth-choice custom-api-key \ --custom-base-url http://127.0.0.1:8080/v1 \ --custom-model-id "cudo528/Darwin-35B-A3B-Opus-mlx-8bit" \ --custom-provider-id mlx-lm \ --custom-compatibility openai \ --custom-text-input \ --accept-risk \ --skip-health
Run OpenClaw
openclaw agent --local --agent main --message "Hello from Hugging Face"
Darwin-35B-A3B-Opus (MLX 8-bit Vision Optimized)
This repository provides an MLX-optimized 8-bit quantized version of FINAL-Bench/Darwin-35B-A3B-Opus, a next-generation Multimodal Vision-Language Model (VLM).
์งํ์ ๋จธ์ง(Evolutionary Merge) ๊ธฐ๋ฒ์ด ์ ์ฉ๋ ๊ณ ์ฑ๋ฅ ๋ชจ๋ธ์ Apple Silicon(Mac) ํ๊ฒฝ์์ ๋ฉ๋ชจ๋ฆฌ ํจ์จ์ ๊ทน๋ํํ์ฌ ๊ตฌ๋ํ ์ ์๋๋ก **8-bit(INT8)**๋ก ์์ํ ๋ฐ ํฌ๋งท ๋ณํ์ ์๋ฃํ ๋ฒ์ ์ ๋๋ค.
๐ Key Features
- 8-bit Quantized: ์๋ณธ ๋ชจ๋ธ์ ๊ฐ๋ ฅํ ์ฑ๋ฅ ์ดํ๋ฅผ ์ต์ํํ๋ฉด์๋, ๋งฅ๋ถ ๋ฑ ๋ก์ปฌ ํ๋์จ์ด ํ๊ฒฝ์์ ๋ถ๋๋ฝ๊ฒ ์ถ๋ก (Inference)ํ ์ ์๋๋ก ์ต์ ์ ๊ฐ์ค์น ์ฉ๋์ผ๋ก ์์ถ๋์์ต๋๋ค.
- Multimodal (Vision-Language): ์ด๋ฏธ์ง์ ํ
์คํธ ์
๋ ฅ์ ๋์์ ์ฒ๋ฆฌํ๋
image-text-to-textํ์ดํ๋ผ์ธ์ ์ง์ํ์ฌ ๋ณต์กํ ์๊ฐ ์๋ฃ ๋ถ์, ๋ํ ํด์, OCR ๋ฑ์ด ๊ฐ๋ฅํฉ๋๋ค. - Advanced Reasoning & CoT: ์ฌ์ธต์ ์ธ ์ฌ๊ณ ํ๋ก์ธ์ค(Thinking)์ Chain-of-Thought(CoT) ๋ฅ๋ ฅ์ ๊ฐ์ถ์ด ๋ ผ๋ฆฌ์ ์ถ๋ก ๋ฅ๋ ฅ์ด ๊ทน๋ํ๋์์ต๋๋ค.
- Massive 262K Context Window: ์ต๋ 262,000 ํ ํฐ์ ์ด์ฅ๋ฌธ ์ปจํ ์คํธ๋ฅผ ์ง์ํ์ฌ ๋ํ ๋ฌธ์ ๋ถ์ ๋ฐ ๊ธด ๋ํ ํ๋ฆ์ ๋๊น ์์ด ์ ์งํฉ๋๋ค.
- Agentic Capabilities: ์ฝ๋ฉ ์์ด์ ํธ(Coding Agent) ๊ธฐ๋ฅ๊ณผ ๊ณ ๋ํ๋ ๋๊ตฌ ํธ์ถ(Tool Calling) ๋ฅ๋ ฅ์ ํ์ฌํ์ฌ ๋ณต์กํ ์ํฌํ๋ก์ฐ๋ฅผ ์์จ์ ์ผ๋ก ์ํํฉ๋๋ค.
๐ ๏ธ Installation
์ด ๋ฉํฐ๋ชจ๋ฌ ๋ชจ๋ธ์ ๊ตฌ๋ํ๋ ค๋ฉด ์ต์ ๋ฒ์ ์ mlx-lm ๋๋ ๋น์ ์ ์ฉ MLX ํจํค์ง๊ฐ ํ์ํฉ๋๋ค.
pip install -U mlx-lm pillow
๐ป How to Use
Python API๋ฅผ ์ด์ฉํ ๋ฉํฐ๋ชจ๋ฌ(์ด๋ฏธ์ง + ํ ์คํธ) ์ถ๋ก ์์
์ด ๋ชจ๋ธ์ ์ด๋ฏธ์ง ์
๋ ฅ์ ์ง์ํ๋ฏ๋ก PIL.Image๋ฅผ ํ์ฉํด ์ง๋ฌธ์ ๋์ง ์ ์์ต๋๋ค.
from mlx_lm import load, generate
from PIL import Image
# ๋ณธ์ธ์ ํ๊น
ํ์ด์ค ์ ์ฅ์ ์์ด๋/์ด๋ฆ์ผ๋ก ์์ ํ์ฌ ์ฌ์ฉํ์ธ์.
model, tokenizer = load("cudo528/Darwin-35B-A3B-Opus-MLX-8bit")
# ๋ถ์ํ ์ด๋ฏธ์ง ๋ก๋
image_path = "sample_chart.png"
image = Image.open(image_path)
# ์ด๋ฏธ์ง์ ํจ๊ป ํ
์คํธ ํ๋กฌํํธ ์์ฑ
prompt = "์ด ์ด๋ฏธ์ง์ ํฌํจ๋ ๋ฐ์ดํฐ ๊ทธ๋ํ๋ฅผ ํ๊ตญ์ด๋ก ์์ธํ ๋ถ์ํ๊ณ ์์ฝํด์ค."
response = generate(
model,
tokenizer,
prompt=prompt,
images=[image], # ๋ฉํฐ๋ชจ๋ฌ ์
๋ ฅ ๋ฐฐ์ด
max_tokens=2048,
verbose=True # ์ฌ์ธต ์ถ๋ก ๊ณผ์ (<thought>)์ ๋ชจ๋ํฐ๋งํ๋ ค๋ฉด True
)
print(response)
๐ Quantization & Lineage
- Original Base Model: FINAL-Bench/Darwin-35B-A3B-Opus
- Quantization Bit: 8-bit (INT8)
- Format: MLX Formatted (
.safetensors) - License: Apache-2.0
- Downloads last month
- 21
8-bit
Model tree for cudo528/Darwin-35B-A3B-Opus-mlx-8bit
Base model
FINAL-Bench/Darwin-35B-A3B-Opus