Instructions to use unsloth/Hunyuan-A13B-Instruct-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use unsloth/Hunyuan-A13B-Instruct-GGUF with Transformers:

# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("unsloth/Hunyuan-A13B-Instruct-GGUF", dtype="auto")

llama-cpp-python

How to use unsloth/Hunyuan-A13B-Instruct-GGUF with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="unsloth/Hunyuan-A13B-Instruct-GGUF",
	filename="BF16/Hunyuan-A13B-Instruct-BF16-00001-of-00004.gguf",
)

llm.create_chat_completion(
	messages = "No input example has been defined for this model task."
)

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use unsloth/Hunyuan-A13B-Instruct-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 unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
llama cli -hf unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
llama cli -hf unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_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 unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
./llama-cli -hf unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_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 unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL

Use Docker

docker model run hf.co/unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL

LM Studio
Jan
Ollama
How to use unsloth/Hunyuan-A13B-Instruct-GGUF with Ollama:
```
ollama run hf.co/unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL
```

Unsloth Studio

How to use unsloth/Hunyuan-A13B-Instruct-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 unsloth/Hunyuan-A13B-Instruct-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 unsloth/Hunyuan-A13B-Instruct-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for unsloth/Hunyuan-A13B-Instruct-GGUF to start chatting

How to use unsloth/Hunyuan-A13B-Instruct-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_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": "unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use unsloth/Hunyuan-A13B-Instruct-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 unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL

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 unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL

Run Hermes

hermes

Atomic Chat new
Docker Model Runner
How to use unsloth/Hunyuan-A13B-Instruct-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL
```

Lemonade

How to use unsloth/Hunyuan-A13B-Instruct-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/Hunyuan-A13B-Instruct-GGUF:UD-Q4_K_XL

Run and chat with the model

lemonade run user.Hunyuan-A13B-Instruct-GGUF-UD-Q4_K_XL

List all available models

lemonade list

Model reply in <answer> Tags

by engrtipusultan - opened Jul 21, 2025

Discussion

engrtipusultan

Jul 21, 2025

In llama.cpp I am using --jinja and --reasoning-format none knobs. Thinking is coming fine in thinking tag in the client but each reply comes within answer tag. For example I asked it about docusauras framework. Reply was:
Docusaurus is an open-source static site generator designed to help teams create and host professional, modern documentation websites.
..............
In summary, Docusaurus simplifies the process of building and maintaining high-quality documentation sites, making it a go-to tool for both technical and non-technical teams.

can we update the jinja template to remove tags ?

engrtipusultan

Aug 22, 2025

•

edited Aug 22, 2025

@shimmyshimmer @danielhanchen Can you kindly fix the chat template. I am unable to stop thinking and also I am getting reply in tags answer.

Hello how are you

🧑‍💻 User

Hello how are you ?

🤖 Assistant

Thinking Process:

Identify the user's input: The user said "Hello how are you ?"
Recognize the intent: This is a standard greeting and a question about my well-being.
Recall my nature: I am an AI, a large language model. I don't have feelings, a physical body, or personal experiences like humans do.
Formulate a polite and informative response:
- Acknowledge the greeting: Start with a friendly "Hello there!"
- Address the question directly but honestly: State that I don't have feelings in the human sense.
- Explain my status: Mention that I'm functioning properly, ready to help, etc. This conveys my operational state effectively.
- Turn the focus back to the user: Ask how I can assist them. This is the primary function and shows helpfulness.
Combine the elements into a coherent response: "Hello there!

As an AI, I don't have feelings in the way humans do, but I'm functioning properly and ready to help. How can I assist you today?"
Review and refine: The response is polite, clear, accurate about my nature, and focuses on being helpful. It directly answers the user's question while maintaining appropriate AI persona.

answer>
Hello there!

As an AI, I don't have feelings in the way humans do, but I'm functioning properly and ready to help. How can I assist you today?
</answer

My llama.cpp settings.

-m /home/tipu/AI/models/unsloth/Hunyuan-A13B-Instruct/Hunyuan-A13B-Instruct-IQ4_XS.gguf --jinja --reasoning-format auto --reasoning-budget 0 -n -1 -ngl 99 --n-cpu-moe 24 --temp 0.5 --top-k 20 --top-p 0.7 --repeat-penalty 1.05 -c 32768 --seed -1 -t 4 -fa --mlock --no-warmup -a Hunyuan-A13B-Non-Thinking --chat-template-kwargs "{\"enable_thinking\": \"False\"}" --no-webui --ubatch-size 128 --batch-size 2048 --cache-type-k q8_0 --cache-type-v q8_0

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment