Instructions to use gary23w/the-veil-12b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use gary23w/the-veil-12b 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 gary23w/the-veil-12b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf gary23w/the-veil-12b:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf gary23w/the-veil-12b:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf gary23w/the-veil-12b: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 gary23w/the-veil-12b:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf gary23w/the-veil-12b: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 gary23w/the-veil-12b:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf gary23w/the-veil-12b:Q4_K_M

Use Docker

docker model run hf.co/gary23w/the-veil-12b:Q4_K_M

LM Studio
Jan

vLLM

How to use gary23w/the-veil-12b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "gary23w/the-veil-12b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "gary23w/the-veil-12b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/gary23w/the-veil-12b:Q4_K_M

Ollama
How to use gary23w/the-veil-12b with Ollama:
```
ollama run hf.co/gary23w/the-veil-12b:Q4_K_M
```
Unsloth Desktop

How to use gary23w/the-veil-12b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf gary23w/the-veil-12b: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": "gary23w/the-veil-12b:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use gary23w/the-veil-12b with Docker Model Runner:
```
docker model run hf.co/gary23w/the-veil-12b:Q4_K_M
```

Lemonade

How to use gary23w/the-veil-12b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull gary23w/the-veil-12b:Q4_K_M

Run and chat with the model

lemonade run user.the-veil-12b-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use gary23w/the-veil-12b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf gary23w/the-veil-12b: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 gary23w/the-veil-12b:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use gary23w/the-veil-12b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf gary23w/the-veil-12b: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 "gary23w/the-veil-12b: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"

the-veil-12b: a Gemma 12B fine-tune that reads its tool list instead of guessing

by gary23w - opened Aug 6

Discussion

gary23w

Owner Aug 6

the-veil-12b is a Gemma 4 12B derivative (GGUF, Apache-2.0, about 7.4 GB for the Q4_K_M) fine-tuned for one failure mode we kept hitting with local models: hand one a JSON tool list and it answers from memory instead of reading it. Give it read_file and it calls Read. Give it web_search and it calls WebSearch. Both get rejected as unknown tools and the turn is lost, which reads as "small models can't do tool use" when the real problem is that the model never looked at the tool list.

What changed

We trained it so the tool array in front of it is the only way to produce a correct answer. Every training example carries a different tool array, sampled across nine schemas plus random subsets of a 58-tool registry, so memorizing any single belt cannot help.

Results on 94 held-out drills, vs the stock model it was tuned from

Drill	Stock	the-veil-12b
General tool use (30)	18	25
Tools never seen in training (24)	18	23
Invented tool names	13	4

Try it

ollama run hf.co/gary23w/the-veil-12b:Q4_K_M

Or llama serve -hf gary23w/the-veil-12b:Q4_K_M, vLLM, LM Studio, Jan, Pi, etc. Full quick starts are on the model card.

It is the first official model for the nl-veil harness (https://github.com/gary23w/nl-veil), which ships 20 tools for chat, 15 for scouts, 8 for assemblers, and bolts on 12 more for browser sessions. Model card with the full details: https://huggingface.co/gary23w/the-veil-12b

Happy to answer questions about the data recipe, the chat template, or the harness. Feedback welcome.

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