Instructions to use VirtuoTuring/chat_noir-24b-v3-gguf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use VirtuoTuring/chat_noir-24b-v3-gguf with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="VirtuoTuring/chat_noir-24b-v3-gguf",
	filename="chatnoir_f16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use VirtuoTuring/chat_noir-24b-v3-gguf with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16
# Run inference directly in the terminal:
llama cli -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16
# Run inference directly in the terminal:
llama cli -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16
# Run inference directly in the terminal:
./llama-cli -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf VirtuoTuring/chat_noir-24b-v3-gguf:F16

Use Docker

docker model run hf.co/VirtuoTuring/chat_noir-24b-v3-gguf:F16

LM Studio
Jan

vLLM

How to use VirtuoTuring/chat_noir-24b-v3-gguf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "VirtuoTuring/chat_noir-24b-v3-gguf"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "VirtuoTuring/chat_noir-24b-v3-gguf",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/VirtuoTuring/chat_noir-24b-v3-gguf:F16

Ollama
How to use VirtuoTuring/chat_noir-24b-v3-gguf with Ollama:
```
ollama run hf.co/VirtuoTuring/chat_noir-24b-v3-gguf:F16
```

Unsloth Studio

How to use VirtuoTuring/chat_noir-24b-v3-gguf 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 VirtuoTuring/chat_noir-24b-v3-gguf 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 VirtuoTuring/chat_noir-24b-v3-gguf to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for VirtuoTuring/chat_noir-24b-v3-gguf to start chatting

Atomic Chat new
Docker Model Runner
How to use VirtuoTuring/chat_noir-24b-v3-gguf with Docker Model Runner:
```
docker model run hf.co/VirtuoTuring/chat_noir-24b-v3-gguf:F16
```

Lemonade

How to use VirtuoTuring/chat_noir-24b-v3-gguf with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull VirtuoTuring/chat_noir-24b-v3-gguf:F16

Run and chat with the model

lemonade run user.chat_noir-24b-v3-gguf-F16

List all available models

lemonade list

chat_noir-24b-v3 — GGUF (Virtuo Turing)

🇬🇧 Overview

v3 ships as GGUF in two builds: F16 and Q4_K_M. Training is more robust with stronger instruction following and better generalisation. Earlier versions leaned to shallow learning and memorisation. v3 is tuned for Portuguese legal comprehension and Portugal-specific topics, including public figures.

What changed in v3

Broader instruction curriculum with reasoning tasks.
Heavier legal mix: CPC, CPP, Constitutional Court, TR/STA/STJ, CJEU, ECtHR.
Reduced overfitting and prompt-echo via objectives and regularisation.
Crisper answers and optional citations when requested.
More robust outside the strict training domain.

Format and compatibility

Files: chat_noir-24b-v3-f16.gguf, chat_noir-24b-v3-q4_k_m.gguf.
Engine: llama.cpp and compatibles (KoboldCpp, text-generation-webui, LM Studio).
Context window ~4k tokens.

Recommended use

PT-PT legal Q&A, short drafting, concise summaries.
Starter config: max_new_tokens=400, temperature=0.2, top_p=0.9, do_sample=true. Tune per task.

Limits

Not long-context. Verify facts when accuracy matters. Not legal advice.

License

Virtuo 1.0. Use and redistribution with attribution to Virtuo Turing – Artificial Intelligence, S.A.

Credits

Developed by Virtuo Turing – Artificial Intelligence, S.A. with Octávio Viana.
Base architecture © Mistral AI (Apache-2.0).

🇵🇹 Resumo

A v3 é lançada em GGUF com dois builds: F16 e Q4_K_M. Traz treino mais robusto, melhor seguimento de instruções e maior generalização. As versões anteriores tendiam à memorização. A v3 foi afinada para compreensão jurídica em PT-PT e para temas ligados a Portugal, incluindo personalidades públicas.

O que mudou no v3

Currículo de instruções mais amplo com tarefas de raciocínio.
Mistura jurídica reforçada: CPC, CPP, Tribunal Constitucional, TR/STA/STJ, TJUE, TEDH.
Menos overfitting e menor eco de prompt por via de objetivos e regularização.
Respostas mais concisas, com citações quando solicitado.
Maior robustez fora do domínio estrito do treino.

Formato e compatibilidade

Ficheiros: chat_noir-24b-v3-f16.gguf, chat_noir-24b-v3-q4_k_m.gguf.
Motor: llama.cpp e compatíveis.
Janela de contexto ~4k tokens.

Uso recomendado

Q&A jurídico PT-PT, minutas curtas e sínteses.
Config inicial: max_new_tokens=400, temperature=0.2, top_p=0.9, do_sample=true. Ajustar por tarefa.

Limitações

Não é long-context. Confirme factos quando crítico. Não constitui aconselhamento jurídico.

Licença

Virtuo 1.0. Uso e redistribuição com atribuição a Virtuo Turing – Artificial Intelligence, S.A.

Créditos

Desenvolvido por Virtuo Turing – Artificial Intelligence, S.A. com Octávio Viana.
Arquitetura base © Mistral AI (Apache-2.0).

Downloads last month: 8

GGUF

Model size

24B params

Architecture

llama

Hardware compatibility

4-bit

16-bit

Model tree for VirtuoTuring/chat_noir-24b-v3-gguf

Base model

mistralai/Mistral-Small-24B-Base-2501

Finetuned

mistralai/Mistral-Small-24B-Instruct-2501

Quantized

(108)

this model