Instructions to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-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 Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL
# Run inference directly in the terminal:
llama cli -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL
# Run inference directly in the terminal:
llama cli -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

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 Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL
# Run inference directly in the terminal:
./llama-cli -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

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 Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL
# Run inference directly in the terminal:
./build/bin/llama-cli -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

Use Docker

docker model run hf.co/Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

LM Studio
Jan

vLLM

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-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": "Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

Ollama
How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with Ollama:
```
ollama run hf.co/Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL
```

Unsloth Studio

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-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 Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-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 Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF to start chatting

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

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": "Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

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 Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

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 "Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL" \
  --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"

Docker Model Runner
How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with Docker Model Runner:
```
docker model run hf.co/Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL
```

Lemonade

How to use Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF:IQ4_NL

Run and chat with the model

lemonade run user.OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF-IQ4_NL

List all available models

lemonade list

OpenThai 2.0 Legal ThaiLLM (Nemotron-3-Nano-30B-A3B) — IQ4_NL GGUF

🇹🇭 Thai Legal AI Model — IQ4_NL Quantized GGUF for llama.cpp

This repository contains the IQ4_NL quantized GGUF version of the OpenThai 2.0 Legal ThaiLLM, based on NVIDIA's Nemotron-3-Nano-30B-A3B architecture.

📊 Model Information

Property	Value
Base Model	iapp/openthai2.0-legal-thaillm-nemotron-3-nano-30b-a3b
Architecture	Nemotron H (Mamba + Attention + MoE Hybrid)
Parameters	~30B total, ~3.5B active per token
Quantization	IQ4_NL (Iterative Optimized 4-bit Non-Linear)
Bits-per-Weight	4.61 BPW
File Size	17 GB
Fallback Tensors	5 of 401 (specialized SSM/Mamba layers)

🔧 Quantization Details

Tool: llama.cpp llama-quantize
Source: F16 GGUF (60 GB, 16.00 BPW)
Quant Method: IQ4_NL — iterative optimized quantization with non-linear quantiles
Quality: Better perplexity than standard Q4_K_M due to iterative optimization
Special Note: 5 tensors used fallback quantization (expected for hybrid Mamba/MoE architecture)

💻 Running the Model

llama.cpp CLI

llama-cli -m "OpenThai 2.0 Legal ThaiLLM (Nemotron-3-Nano-30B-A3B)_IQ4_NL.gguf" \
  -n 1024 \
  -t 8 \
  -p "คำถามทางกฎหมายของคุณ..."

llama-server (OpenAI-compatible API)

llama-server -m "OpenThai 2.0 Legal ThaiLLM (Nemotron-3-Nano-30B-A3B)_IQ4_NL.gguf" \
  --host 0.0.0.0 \
  --port 8080 \
  --ctx-size 4096

Python (llama-cpp-python)

from llama_cpp import Llama

model = Llama.from_pretrained(
    repo_id="Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF",
    filename="OpenThai 2.0 Legal ThaiLLM (Nemotron-3-Nano-30B-A3B)_IQ4_NL.gguf",
    n_ctx=4096,
    n_gpu_layers=-1,  # Offload all layers to GPU
)

output = model("คำถาม: กฎหมายไทยเกี่ยวกับ...", max_tokens=512)
print(output["choices"][0]["text"])

🖥️ Hardware Requirements

Component	Minimum	Recommended
VRAM	16 GB	24 GB (RTX 3090)
RAM	32 GB	64 GB
Storage	20 GB free	30 GB free

📝 License

This model inherits the license from the base model: Apache 2.0

🔗 Links

Original Model: iapp/openthai2.0-legal-thaillm-nemotron-3-nano-30b-a3b
llama.cpp: github.com/ggerganov/llama.cpp
NVIDIA Nemotron: developer.nvidia.com/nemotron

👤 Author

Quantized and shared by @Naypa

If you find this model useful, please ⭐ star the original repo and consider citing the Nemotron architecture paper.

Downloads last month: 299

GGUF

Model size

32B params

Architecture

nemotron_h_moe

Hardware compatibility

4-bit

Model tree for Naypa/OpenThai-2.0-Legal-ThaiLLM-Nemotron-3-Nano-30B-A3B-IQ4_NL-GGUF

Base model

nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16

Finetuned

iapp/openthai2.0-legal-thaillm-nemotron-3-nano-30b-a3b

Quantized

(4)

this model