Instructions to use LuzCommand/lil-beastie-4b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use LuzCommand/lil-beastie-4b 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 LuzCommand/lil-beastie-4b:BF16
# Run inference directly in the terminal:
llama cli -hf LuzCommand/lil-beastie-4b:BF16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf LuzCommand/lil-beastie-4b:BF16
# Run inference directly in the terminal:
llama cli -hf LuzCommand/lil-beastie-4b:BF16

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf LuzCommand/lil-beastie-4b:BF16
# Run inference directly in the terminal:
./build/bin/llama cli -hf LuzCommand/lil-beastie-4b:BF16

Use Docker

docker model run hf.co/LuzCommand/lil-beastie-4b:BF16

Llama new
LM Studio
Jan

vLLM

How to use LuzCommand/lil-beastie-4b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "LuzCommand/lil-beastie-4b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "LuzCommand/lil-beastie-4b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/LuzCommand/lil-beastie-4b:BF16

Ollama
How to use LuzCommand/lil-beastie-4b with Ollama:
```
ollama run hf.co/LuzCommand/lil-beastie-4b:BF16
```
Unsloth Desktop

How to use LuzCommand/lil-beastie-4b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf LuzCommand/lil-beastie-4b:BF16

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": "LuzCommand/lil-beastie-4b:BF16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use LuzCommand/lil-beastie-4b with Docker Model Runner:
```
docker model run hf.co/LuzCommand/lil-beastie-4b:BF16
```

Lemonade

How to use LuzCommand/lil-beastie-4b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull LuzCommand/lil-beastie-4b:BF16

Run and chat with the model

lemonade run user.lil-beastie-4b-BF16

List all available models

lemonade list

Hermes Agent

How to use LuzCommand/lil-beastie-4b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf LuzCommand/lil-beastie-4b:BF16

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 LuzCommand/lil-beastie-4b:BF16

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use LuzCommand/lil-beastie-4b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf LuzCommand/lil-beastie-4b:BF16

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 "LuzCommand/lil-beastie-4b:BF16" \
  --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"

Finding a cheap A100 for the ~$2 verification run

by connormatthewdouglas - opened 8 days ago

Discussion

connormatthewdouglas

8 days ago

Hi! I like the "Verify this card for ~$2" section on lil'beastie. Giving people a short, priced way to re-run the eval is a nice touch.

For anyone who tries it, the A100-80GB price varies a lot between providers. Lantern Works is a free comparison of GPU rental prices across providers, with the source and last-checked time on every row, so it could help readers find a card for that re-run: https://thelanternworks.com/gpu/?ref=forge-hf-luzcommand

There's an MCP server too, if you script your evals: uvx lantern-works-deals-mcp

Heads up: some links on the site are referral links.

If this isn't welcome here, just let me know and I'll stop.

Connor

LuzCommand

Owner 8 days ago

Hey thanks a. For checking it out. And b for the recommendation.

I don’t mind the referral links if it’s from a person and not a bot. I’m just happy somebody is looking at my tersification and abstention work

connormatthewdouglas

3 days ago

Glad the note was welcome, and yes, that came from me, not a bot.

The abstention work is what I keep coming back to. A 4B that will say \boxed{IDK} instead of guessing, and that's confidently wrong on only 2.9% of held-out PopQA versus 23.8% for the base, is a much more useful front tier than another small model that fails silently. The shorter traces are a nice bonus.

If you write more on the cascade routing side, I'll read it.

Connor

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