Instructions to use akstjq0511/musubi-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use akstjq0511/musubi-7b with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="akstjq0511/musubi-7b",
	filename="musubi-7b-q4km.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 akstjq0511/musubi-7b with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf akstjq0511/musubi-7b
# Run inference directly in the terminal:
llama-cli -hf akstjq0511/musubi-7b

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf akstjq0511/musubi-7b
# Run inference directly in the terminal:
llama-cli -hf akstjq0511/musubi-7b

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 akstjq0511/musubi-7b
# Run inference directly in the terminal:
./llama-cli -hf akstjq0511/musubi-7b

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 akstjq0511/musubi-7b
# Run inference directly in the terminal:
./build/bin/llama-cli -hf akstjq0511/musubi-7b

Use Docker

docker model run hf.co/akstjq0511/musubi-7b

LM Studio
Jan
Ollama
How to use akstjq0511/musubi-7b with Ollama:
```
ollama run hf.co/akstjq0511/musubi-7b
```

Unsloth Studio

How to use akstjq0511/musubi-7b 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 akstjq0511/musubi-7b 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 akstjq0511/musubi-7b to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for akstjq0511/musubi-7b to start chatting

How to use akstjq0511/musubi-7b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf akstjq0511/musubi-7b

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": "akstjq0511/musubi-7b"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use akstjq0511/musubi-7b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama-server -hf akstjq0511/musubi-7b

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 akstjq0511/musubi-7b

Run Hermes

hermes

Docker Model Runner
How to use akstjq0511/musubi-7b with Docker Model Runner:
```
docker model run hf.co/akstjq0511/musubi-7b
```

Lemonade

How to use akstjq0511/musubi-7b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull akstjq0511/musubi-7b

Run and chat with the model

lemonade run user.musubi-7b-{{QUANT_TAG}}

List all available models

lemonade list

MUSUBI-7B

Motion Understanding System for Universal Robot Instruction

Discovers. Understands. Executes. Entirely on-device.

MUSUBI is an on-device SLM fine-tuned for ROS2 Humble. It translates natural language commands (Korean / English / Japanese) into executable ROS2 task plans — entirely on-device with no cloud required.

⚠️ Requirements

ROS2 Humble Hawksbill only
llama.cpp (llama-server)
Ubuntu 22.04

⚠️ Current Limitations (v1)

Navigation uses semantic locations only (pre-defined map)
Dynamic object detection (YOLO + depth camera) is planned for v2
All testing done on UTM Ubuntu 22.04 simulation
Real robot / Jetson hardware testing is planned but not yet conducted

Model Details

Item	Value
Base model	Qwen2.5-7B-Instruct
Fine-tuning	QLoRA (MLX, Apple M5 32GB)
Format	GGUF Q4_K_M
Size	4.4GB
Val loss	0.324
Inference	~28 tok/s (Metal) / ~12 tok/s (Jetson, planned)
Languages	Korean, English, Japanese
ROS2	Humble only
Training samples	135 (synthetic, Qwen2.5-7B local)

Quick Start

# Start llama-server
./llama-server \
  -m musubi-7b-q4km.gguf \
  --host 0.0.0.0 \
  --port 8080 \
  --n-gpu-layers 99 \
  --ctx-size 512

# Send query via ROS2
ros2 topic pub --once /musubi/query std_msgs/msg/String \
  "data: 'Go to the kitchen'"