Instructions to use shafire/OpenZero-TerraPilot-3.8B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use shafire/OpenZero-TerraPilot-3.8B-GGUF with Transformers:

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("shafire/OpenZero-TerraPilot-3.8B-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use shafire/OpenZero-TerraPilot-3.8B-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 shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf shafire/OpenZero-TerraPilot-3.8B-GGUF: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 shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf shafire/OpenZero-TerraPilot-3.8B-GGUF: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 shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M

Llama new
LM Studio
Jan
Ollama
How to use shafire/OpenZero-TerraPilot-3.8B-GGUF with Ollama:
```
ollama run hf.co/shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M
```
Unsloth Desktop

How to use shafire/OpenZero-TerraPilot-3.8B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf shafire/OpenZero-TerraPilot-3.8B-GGUF: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": "shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use shafire/OpenZero-TerraPilot-3.8B-GGUF with Docker Model Runner:
```
docker model run hf.co/shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M
```

Lemonade

How to use shafire/OpenZero-TerraPilot-3.8B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.OpenZero-TerraPilot-3.8B-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use shafire/OpenZero-TerraPilot-3.8B-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 shafire/OpenZero-TerraPilot-3.8B-GGUF: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 shafire/OpenZero-TerraPilot-3.8B-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use shafire/OpenZero-TerraPilot-3.8B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf shafire/OpenZero-TerraPilot-3.8B-GGUF: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 "shafire/OpenZero-TerraPilot-3.8B-GGUF: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"

OpenZero TerraPilot 3.8B GGUF

Private research release — not a certified robot controller or deployment-ready autonomy system.

OpenZero TerraPilot 3.8B is an experimental mission-level ground-robotics proposal model derived from the pinned microsoft/Phi-4-mini-instruct revision cfbefacb99257ffa30c83adab238a50856ac3083.

It is designed to emit typed openzero.motion.v1 mission proposals for ROS 2/Nav2 integration research. Deterministic controllers, configured limits, collision monitoring, emergency stop and operator authority remain outside the model and retain control.

Verified artifact

Field	Value
File	`OpenZero-TerraPilot-3.8B-Q4_K_M.gguf`
Quantization	Q4_K_M
Bytes	2,493,840,128
SHA-256	`8b40776a2a457a27b74466a90e664ee854e66e468f157d116dd9deaac5513dfe`
llama.cpp	b10451, commit `10bf611e533d81f739128304991c5e133c6aebd8`
Runtime gate	Passed, 8 generated tokens, 5.8 tokens/s

Training and provenance

Two-step finite QLoRA smoke run on a Tesla T4.
32 deterministic locally authored training cases.
6 disjoint validation cases.
Training loss: 4.98442006111145.
Adapter SHA-256: fac55cf3152760cb4787f60d801a1e02424d0f4aaaf49c6d0002faacb814ca88.
Excluded sources: Fusion, teacher outputs, locked evaluation, rejected data, Ministral data and failed Gemma-31B checkpoints.

This very small smoke dataset demonstrates pipeline integrity, not broad robotics competence or superiority over the base model.

Intended research uses

ROS 2/Nav2 mission-envelope experiments
waypoint and relative-motion proposal validation
collision/fault response schema testing
infrastructure inspection and benign logistics simulations
simulator, SITL and later controlled HITL commissioning research

Boundaries

TerraPilot does not directly command motors or actuators. It is not designed for weapons, harmful payloads, person targeting or pursuit, restraint, evasion, interference, unauthorised control or safety-controller bypass. Every proposal requires deterministic validation and an independently enforceable authority boundary.

Status

experimental GGUF. No MOD, UKRI, OpenAI, Microsoft or other institutional endorsement is claimed.

Downloads last month: 85

GGUF

Model size

4B params

Architecture

phi3

Hardware compatibility

4-bit

Video Preview

Robotics

Model tree for shafire/OpenZero-TerraPilot-3.8B-GGUF

Base model

microsoft/Phi-4-mini-instruct

Quantized

(201)

this model