Instructions to use NeveAI/Neve-Strata-S2-4B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use NeveAI/Neve-Strata-S2-4B-GGUF with Transformers:

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

pipe = pipeline("text-generation", model="NeveAI/Neve-Strata-S2-4B-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("NeveAI/Neve-Strata-S2-4B-GGUF", dtype="auto")

llama-cpp-python

How to use NeveAI/Neve-Strata-S2-4B-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="NeveAI/Neve-Strata-S2-4B-GGUF",
	filename="Neve-Strata-S2-4B-Q8_K_XL.gguf",
)

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

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use NeveAI/Neve-Strata-S2-4B-GGUF with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL
# Run inference directly in the terminal:
llama-cli -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL
# Run inference directly in the terminal:
llama-cli -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

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 NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL
# Run inference directly in the terminal:
./llama-cli -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

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 NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL
# Run inference directly in the terminal:
./build/bin/llama-cli -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

Use Docker

docker model run hf.co/NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

LM Studio
Jan

vLLM

How to use NeveAI/Neve-Strata-S2-4B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "NeveAI/Neve-Strata-S2-4B-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": "NeveAI/Neve-Strata-S2-4B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

SGLang

How to use NeveAI/Neve-Strata-S2-4B-GGUF 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 "NeveAI/Neve-Strata-S2-4B-GGUF" \
    --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": "NeveAI/Neve-Strata-S2-4B-GGUF",
		"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 "NeveAI/Neve-Strata-S2-4B-GGUF" \
        --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": "NeveAI/Neve-Strata-S2-4B-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use NeveAI/Neve-Strata-S2-4B-GGUF with Ollama:
```
ollama run hf.co/NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL
```

Unsloth Studio new

How to use NeveAI/Neve-Strata-S2-4B-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 NeveAI/Neve-Strata-S2-4B-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 NeveAI/Neve-Strata-S2-4B-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for NeveAI/Neve-Strata-S2-4B-GGUF to start chatting

Pi new

How to use NeveAI/Neve-Strata-S2-4B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Neve-Strata-S2-4B-GGUF"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use NeveAI/Neve-Strata-S2-4B-GGUF with Docker Model Runner:
```
docker model run hf.co/NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL
```

Lemonade

How to use NeveAI/Neve-Strata-S2-4B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull NeveAI/Neve-Strata-S2-4B-GGUF:Q8_K_XL

Run and chat with the model

lemonade run user.Neve-Strata-S2-4B-GGUF-Q8_K_XL

List all available models

lemonade list

Neve-Strata-S2-4B-GGUF

File size: 3,118 Bytes

f0debc8
 
 
 
283ebd7
f0debc8
 
 
 
 
 
 
 
5d4c6b7
 
 
f0debc8
 
 
5d4c6b7
f0debc8
5d4c6b7
f0debc8
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
559958f
 
 
 
 
 
 
 
f0debc8

---
library_name: transformers
license: apache-2.0
license_link: https://huggingface.co/Qwen/Qwen3.5-4B/blob/main/LICENSE
pipeline_tag: text-generation
base_model:
- Qwen/Qwen3.5-4B
tags:
  - NeveAI
  - Neve
  - StrataS
---

<div align="center">
  <img src="https://cdn-uploads.huggingface.co/production/uploads/68a3ba234a7dfca33d72eee2/BG71IO9zlNcw4eTRYKZzO.png" width="50%">
</div>

<h1 align="center">Neve-Strata-S2-4B-GGUF</h1>

<div align="center">
  <a href="https://github.com/NeveIA">
    <img src="https://cdn-uploads.huggingface.co/production/uploads/68a3ba234a7dfca33d72eee2/KQa7-ajynUAhTS-kxNtYT.png" width="20%" alt="NeveAI GitHub">
  </a>
</div>

## Introdução

O **Neve Strata S2** é um modelo de linguagem de última geração focado em **programação e raciocínio para execução em escala**. Esta versão em formato GGUF foi otimizada pela NeveAI para oferecer o equilíbrio ideal entre precisão lógica e eficiência computacional.

---

## Destaques do Modelo

Este modelo foi desenvolvido para uso geral e execução de tarefas diversas, focando em:

*   **Unified Multimodal Understanding:** Treinamento com fusão antecipada de tokens multimodais, garantindo forte desempenho em tarefas de texto e compreensão visual.
*   **Arquitetura Híbrida Eficiente:** Combinação de Gated Delta Networks com Mixture-of-Experts, proporcionando alta performance com baixa latência.
*   **Raciocínio e Generalização:** Otimizado com técnicas avançadas de reinforcement learning para lidar com tarefas complexas e cenários do mundo real.
*   **Cobertura Multilíngue Global:** Suporte expandido para múltiplos idiomas, garantindo aplicação ampla em diferentes contextos culturais e linguísticos.

## Benchmark de Performance

O Neve Strata S2 apresenta desempenho sólido em benchmarks de conhecimento, raciocínio e tarefas gerais:

| Categoria | Benchmark | Neve Strata S2 | Qwen3.5-4B |
| :--- | :--- | :---: | :---: |
| **Knowledge** | MMLU-Pro | **82.5** | 79.1 |
| **Knowledge** | MMLU-Redux | **91.1** | 88.8 |
| **Reasoning** | GPQA Diamond | **81.7** | 76.2 |
| **Instruction** | IFEval | **91.5** | 89.8 |
| **Long Context** | LongBench v2 | **55.2** | 50.0 |
| **Agent / Tool Use** | TAU2-Bench | 79.1 | **79.9** |

---

## Detalhes da Arquitetura

- **Arquitetura:** Gated DeltaNet + Mixture of Experts (MoE).
- **Parâmetros:** ~4B parâmetros.
- **Janela de Contexto:** 262.144 tokens nativos (extensível até ~1M).
- **Camadas:** 32 camadas com estrutura híbrida intercalando DeltaNet e Attention.
- **Multimodalidade:** Suporte a texto e visão com encoder integrado.

## Como utilizar (GGUF)

Este modelo é compatível com `llama.cpp`, `Ollama`, `LM Studio` e outras ferramentas que suportam o formato GGUF. 
Foco direcionado ao uso do modelo na plataforma autoral da organização [NeveAI](https://github.com/Etamus/NeveAI)

## Licença

Este repositório e os pesos do modelo estão licenciados sob a [Licença Apache 2.0](LICENSE).

## Contato

Se tiver qualquer dúvida, por favor, levante um issue ou entre em contato conosco em [NeveIA](https://github.com/NeveIA).