Instructions to use saijeevanp/orbit-scam-defense with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use saijeevanp/orbit-scam-defense 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 saijeevanp/orbit-scam-defense:Q6_K
# Run inference directly in the terminal:
llama cli -hf saijeevanp/orbit-scam-defense:Q6_K

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf saijeevanp/orbit-scam-defense:Q6_K
# Run inference directly in the terminal:
llama cli -hf saijeevanp/orbit-scam-defense:Q6_K

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 saijeevanp/orbit-scam-defense:Q6_K
# Run inference directly in the terminal:
./llama-cli -hf saijeevanp/orbit-scam-defense:Q6_K

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 saijeevanp/orbit-scam-defense:Q6_K
# Run inference directly in the terminal:
./build/bin/llama-cli -hf saijeevanp/orbit-scam-defense:Q6_K

Use Docker

docker model run hf.co/saijeevanp/orbit-scam-defense:Q6_K

LM Studio
Jan

vLLM

How to use saijeevanp/orbit-scam-defense with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/saijeevanp/orbit-scam-defense:Q6_K

Ollama
How to use saijeevanp/orbit-scam-defense with Ollama:
```
ollama run hf.co/saijeevanp/orbit-scam-defense:Q6_K
```
Unsloth Desktop

How to use saijeevanp/orbit-scam-defense with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf saijeevanp/orbit-scam-defense:Q6_K

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": "saijeevanp/orbit-scam-defense:Q6_K"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use saijeevanp/orbit-scam-defense with Docker Model Runner:
```
docker model run hf.co/saijeevanp/orbit-scam-defense:Q6_K
```

Lemonade

How to use saijeevanp/orbit-scam-defense with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull saijeevanp/orbit-scam-defense:Q6_K

Run and chat with the model

lemonade run user.orbit-scam-defense-Q6_K

List all available models

lemonade list

Hermes Agent

How to use saijeevanp/orbit-scam-defense with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf saijeevanp/orbit-scam-defense:Q6_K

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 saijeevanp/orbit-scam-defense:Q6_K

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use saijeevanp/orbit-scam-defense with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf saijeevanp/orbit-scam-defense:Q6_K

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 "saijeevanp/orbit-scam-defense:Q6_K" \
  --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"

Orbit: on-device scam defense for the people you love

This is the model file used by the Orbit Android app. It is a 1-billion-parameter open model (MiniCPM5-1B) fine-tuned with a LoRA adapter to read a suspicious message and answer with a strict JSON verdict: risk level, who the sender pretends to be, how they pressure you, what they want, what could happen, and the one safe next step.

The Orbit app downloads this file once and then runs it entirely on the phone with llama.cpp. Messages, verdicts, history and labels never leave the device.

Files

File	Size	Use
`orbit-q6_k.gguf`	892 MB	Ships in the app. 23/24 exact on a stratified on-device sample, zero dangerous under-calls.
`adapter/`	86 MB	The LoRA adapter (PEFT, r=32, alpha=64) on `openbmb/MiniCPM5-1B`, for the web app's Transformers backend: `PeftModel.from_pretrained(base, "saijeevanp/orbit-scam-defense", subfolder="adapter")`.

Prompt contract

The app sends a fixed system prompt and the message as the user turn. Output is a single JSON object. The app validates it, repairs truncation, and runs a deterministic rule layer that can overrule the model toward caution. Do not expect useful output without the app's system prompt.

Limits

Trained on English scam messages with India-specific patterns (KYC, UPI, digital arrest, WhatsApp job offers, prize claims). Explanation quality drops on patterns it never saw; the app's rule layer supplies the explanation in those cases.
Scam-type labels can drift; the risk level is what the app's guard protects.
Not a substitute for calling your bank.

Team

AI Pirates, iQOO Hackathon 2026, FinTech and Commerce track.

Downloads last month: 46

GGUF

Model size

1B params

Architecture

llama

Hardware compatibility

6-bit

Model tree for saijeevanp/orbit-scam-defense

Base model

openbmb/MiniCPM5-1B

Quantized

(100)

this model