Instructions to use muverqqw/Noir-14B-Starlight with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use muverqqw/Noir-14B-Starlight with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="muverqqw/Noir-14B-Starlight")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("muverqqw/Noir-14B-Starlight")
model = AutoModelForCausalLM.from_pretrained("muverqqw/Noir-14B-Starlight", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use muverqqw/Noir-14B-Starlight with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "muverqqw/Noir-14B-Starlight"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "muverqqw/Noir-14B-Starlight",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/muverqqw/Noir-14B-Starlight

SGLang

How to use muverqqw/Noir-14B-Starlight with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "muverqqw/Noir-14B-Starlight" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "muverqqw/Noir-14B-Starlight",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "muverqqw/Noir-14B-Starlight" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "muverqqw/Noir-14B-Starlight",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Unsloth Studio

How to use muverqqw/Noir-14B-Starlight with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for muverqqw/Noir-14B-Starlight to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for muverqqw/Noir-14B-Starlight to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for muverqqw/Noir-14B-Starlight to start chatting

Load model with FastModel

pip install unsloth
from unsloth import FastModel
model, tokenizer = FastModel.from_pretrained(
    model_name="muverqqw/Noir-14B-Starlight",
    max_seq_length=2048,
)

Docker Model Runner
How to use muverqqw/Noir-14B-Starlight with Docker Model Runner:
```
docker model run hf.co/muverqqw/Noir-14B-Starlight
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

🌌 Noir-14B-Starlight

Noir-14B-Starlight is the crown jewel of the Noir family. As the largest and most capable model in the series, Starlight is designed for users who refuse to compromise between depth of knowledge and reasoning precision.

By leveraging its 14-billion parameter architecture, Starlight bridges the gap between mid-range efficiency and frontier-level intelligence, delivering exceptional performance in complex problem-solving and nuanced linguistic tasks.

✨ Key Enhancements

The "Starlight" Reasoning: Advanced fine-tuning focused on multi-step logic and reducing factual hallucinations.
Architectural Superiority: The 14B parameter count provides a significant "intelligence jump" over the Ultra and Lightning versions, especially in creative writing and coding.
Long-Context Stability: Optimized to maintain coherence and follow strict formatting instructions during long-form generation.
Nuanced Multilingualism: Enhanced support for complex grammar and cultural nuances across multiple languages.

📊 The Noir Hierarchy

Model	Parameters	Role	Key Strength
Noir-Lightning	0.5B	The Pocket Assistant	Ultra-fast, runs on anything
Noir-Mini	1.5B	The Balanced Thinker	High speed with solid grammar
Noir-Standard	3B	The Versatile Workhorse	65% GSM8K, perfect for 8GB VRAM
Noir-Ultra	7B	The Reasoning Master	91% SciQ & 84% Math
Noir-Starlight	14B	The Galactic Intelligence	Deep logic & Expert-level STEM

🚀 Implementation

Noir-14B-Starlight is compatible with the transformers library. For optimal performance, we recommend using 4-bit or 8-bit quantization if VRAM is limited.

from transformers import AutoModelForCausalLM, AutoTokenizer
import torch

model_id = "muverqqw/Noir-14B-Starlight"

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
    trust_remote_code=True
)

messages = [
    {"role": "system", "content": "You are Starlight, the most advanced AI of the Noir series."},
    {"role": "user", "content": "Write a complex Python script for an asynchronous web scraper."}
]

inputs = tokenizer.apply_chat_template(messages, add_generation_prompt=True, return_tensors="pt").to("cuda")
outputs = model.generate(inputs, max_new_tokens=1024, do_sample=True, temperature=0.7)

print(tokenizer.decode(outputs[0], skip_special_tokens=True))

⚖️ Technical Specifications

Parameters: 14 Billion
Format: Safetensors
Recommended Temperature: 0.5 - 0.8 (depending on the task)

🛡 Limitations & Ethical Note

While Starlight is our most robust model, it may still produce incorrect information in highly niche technical fields. Always verify critical data. This model is intended for research and creative assistance.

Developed with ❤️ by IceL1ghtning

Downloads last month: 8

Safetensors

Model size

15B params

Tensor type

BF16

Model tree for muverqqw/Noir-14B-Starlight

Unable to build the model tree, the base model loops to the model itself. Learn more.

Collection including muverqqw/Noir-14B-Starlight

Noir

Collection

Our text model is woven from miracles, magic and dreams. • 8 items • Updated Apr 30 • 2