Instructions to use N14QM/Vietnamese-4B-Chat with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use N14QM/Vietnamese-4B-Chat with Transformers:

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

pipe = pipeline("text-generation", model="N14QM/Vietnamese-4B-Chat")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("N14QM/Vietnamese-4B-Chat", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use N14QM/Vietnamese-4B-Chat 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 N14QM/Vietnamese-4B-Chat:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf N14QM/Vietnamese-4B-Chat:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

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 N14QM/Vietnamese-4B-Chat:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

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 N14QM/Vietnamese-4B-Chat:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

Use Docker

docker model run hf.co/N14QM/Vietnamese-4B-Chat:Q4_K_M

LM Studio
Jan

vLLM

How to use N14QM/Vietnamese-4B-Chat with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "N14QM/Vietnamese-4B-Chat"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "N14QM/Vietnamese-4B-Chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/N14QM/Vietnamese-4B-Chat:Q4_K_M

SGLang

How to use N14QM/Vietnamese-4B-Chat 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 "N14QM/Vietnamese-4B-Chat" \
    --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": "N14QM/Vietnamese-4B-Chat",
		"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 "N14QM/Vietnamese-4B-Chat" \
        --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": "N14QM/Vietnamese-4B-Chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use N14QM/Vietnamese-4B-Chat with Ollama:
```
ollama run hf.co/N14QM/Vietnamese-4B-Chat:Q4_K_M
```
Unsloth Desktop

How to use N14QM/Vietnamese-4B-Chat with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "N14QM/Vietnamese-4B-Chat:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use N14QM/Vietnamese-4B-Chat with Docker Model Runner:
```
docker model run hf.co/N14QM/Vietnamese-4B-Chat:Q4_K_M
```

Lemonade

How to use N14QM/Vietnamese-4B-Chat with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull N14QM/Vietnamese-4B-Chat:Q4_K_M

Run and chat with the model

lemonade run user.Vietnamese-4B-Chat-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use N14QM/Vietnamese-4B-Chat with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

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 N14QM/Vietnamese-4B-Chat:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use N14QM/Vietnamese-4B-Chat with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf N14QM/Vietnamese-4B-Chat:Q4_K_M

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 "N14QM/Vietnamese-4B-Chat:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

📊 Tại sao tỷ lệ 0.2 / 0.3 / 0.5?

Model	Vai trò	Lý do tỷ lệ
DeepSeek-V4-Pro (0.2)	Nền tảng reasoning	Giữ logic gốc, không pha loãng
KIMI-Distill (0.3)	Tăng cường suy luận	Boost chain-of-thought
Vietnamese-SFT (0.5)	Ngôn ngữ chính	Đảm bảo tiếng Việt tự nhiên

Tỷ lệ này được chọn qua thử nghiệm — cân bằng giữa kiến thức (DeepSeek) và ngôn ngữ (Vietnamese).

📊 Kết quả Benchmark

V-Bench (VinUni) — Public Test v2026.03.28

Tổng số câu: 5,141/9,141 được chấm

Domain	Số đúng	Tổng	Điểm
TỔNG	—	—	32.49 / 100
Philosophy	154	263	58.56%
Computer Science	109	188	57.98%
Dialect	265	562	47.15%
Laws	90	191	47.12%
Chemistry	107	334	32.04%
Psychology	247	798	30.95%
Medicine	136	490	27.76%
Physics	36	147	24.49%
Literature	140	601	23.29%
Mathematics	24	125	19.20%
Agentic	0	1000	0.00%

📈 So sánh với các model khác

Model	Params	V-Bench	Ghi chú
Vietnamese-4B-Chat	4B	32.49	Model này
DeepSeek-V4-Pro-Qwen3.5	4B	29.14	Base gốc
Hunyuan-4B	4B	17.24	So sánh
Random baseline	—	25.00	Ngẫu nhiên
Qwen3-235B	235B	~55	Model lớn
Gemini 2.5 Flash	?	58.8	Closed API
V-LLM v1 (Vin)	~30B	68.00	SOTA VI

Kết luận: Mô hình vượt base gốc +3.35 điểm và vượt random +7.49 điểm.

🎯 Điểm mạnh / yếu

✅ Mạnh:

Triết học, Computer Science (58%)
Luật, Phương ngữ tiếng Việt (47%)
Trò chuyện tiếng Việt tự nhiên
Code với comment VI
Tốc độ nhanh (~5 câu/s, không cần thinking)

❌ Yếu:

Toán (19.20%) — dưới random, do tắt thinking
Vật lý (24.49%) — kế thừa từ base
Agentic (0%) — cần model 30B+
Không hỗ trợ hình ảnh (text-only)

🚀 Cách sử dụng

1️⃣ Với llama.cpp (khuyến nghị)

# Cài đặt llama.cpp
git clone https://github.com/ggerganov/llama.cpp
cd llama.cpp
cmake -B build
cmake --build build --config Release -j

# Chạy chat
./build/bin/llama-cli \
  -m Vietnamese-4B-Chat-Q4_K_M.gguf \
  -p "Chào bạn, hôm nay thế nào?" \
  -n 256 \
  -c 2048 \
  --temp 0.7 \
  --top-p 0.9

# Chế độ tương tác
./build/bin/llama-cli \
  -m Vietnamese-4B-Chat-Q4_K_M.gguf \
  -i --color

Downloads last month: 615

GGUF

Model size

4B params

Architecture

qwen35

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

Model tree for N14QM/Vietnamese-4B-Chat

Jackrong/DeepSeek-V4-Pro-Qwen3.5-4B

Qwen/Qwen3.5-4B

clzoro/Qwen3.5-4B-KIMI-Distill

Merge model

this model

Quantizations

1 model