Instructions to use eddyejembi/vital-med-tiny with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use eddyejembi/vital-med-tiny 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 eddyejembi/vital-med-tiny:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf eddyejembi/vital-med-tiny:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf eddyejembi/vital-med-tiny:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf eddyejembi/vital-med-tiny: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 eddyejembi/vital-med-tiny:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf eddyejembi/vital-med-tiny: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 eddyejembi/vital-med-tiny:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf eddyejembi/vital-med-tiny:Q4_K_M

Use Docker

docker model run hf.co/eddyejembi/vital-med-tiny:Q4_K_M

LM Studio
Jan

vLLM

How to use eddyejembi/vital-med-tiny with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/eddyejembi/vital-med-tiny:Q4_K_M

Ollama
How to use eddyejembi/vital-med-tiny with Ollama:
```
ollama run hf.co/eddyejembi/vital-med-tiny:Q4_K_M
```

Unsloth Studio

How to use eddyejembi/vital-med-tiny 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 eddyejembi/vital-med-tiny 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 eddyejembi/vital-med-tiny to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for eddyejembi/vital-med-tiny to start chatting

How to use eddyejembi/vital-med-tiny with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf eddyejembi/vital-med-tiny:Q4_K_M

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": "eddyejembi/vital-med-tiny:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use eddyejembi/vital-med-tiny with Docker Model Runner:
```
docker model run hf.co/eddyejembi/vital-med-tiny:Q4_K_M
```

Lemonade

How to use eddyejembi/vital-med-tiny with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull eddyejembi/vital-med-tiny:Q4_K_M

Run and chat with the model

lemonade run user.vital-med-tiny-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use eddyejembi/vital-med-tiny with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf eddyejembi/vital-med-tiny: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 eddyejembi/vital-med-tiny:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use eddyejembi/vital-med-tiny with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf eddyejembi/vital-med-tiny: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 "eddyejembi/vital-med-tiny: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"

Vital-Med Tiny

Vital-Med Tiny is a Qwen3.5-2B based GGUF model fine-tuned for offline healthcare and wellness guidance in African low-resource contexts.

The selected model is intended for CPU-only llama.cpp inference on ordinary laptops. It was built for the Africa Deep Tech Challenge 2026 Laptop LLM track.

Files

File	Description
`vital-med-Q4_K_M.gguf`	Selected GGUF submission artifact

Intended Use

Vital-Med Tiny provides educational health-information and wellness guidance:

symptom triage framing
red-flag reminders
malaria-aware patient education
child fever and dehydration guidance
blood-pressure and wellness coaching
desk-worker hydration and ergonomics

It is not a doctor, diagnostic system, or emergency service.

Runtime

llama-cli -m vital-med-Q4_K_M.gguf -p "My 3-year-old in Lagos has had fever for 2 days and is drinking poorly. What should I do now?"

If your llama.cpp build supports chat mode:

llama-cli -m vital-med-Q4_K_M.gguf -cnv

Model Details

Field	Value
Base	Qwen3.5-2B
Fine-tuning	bf16 LoRA SFT
Runtime	`llama.cpp`
Format	GGUF
Quantization	`Q4_K_M`
Primary language	English
Domain	Healthcare / medical guidance / wellness

Local Profiler Results

Measured on participant laptop with adtc-profiler 0.1.0:

Metric	Result
Generation throughput	10.41 tokens/s
First-token latency	13.24 s
Peak RSS	2.01 GB
Steady RSS	1.93 GB
`arc_easy` proxy	0.68 acc_norm, 50 samples
Thermal throttling	No

The model stays comfortably under the ADTC 7 GB RAM ceiling.

Model Selection

Three post-training candidates were evaluated:

Candidate	Decision
Original SFT Q4_K_M	Selected
Correction SFT	Rejected
Tiny DPO pass	Rejected

The correction and DPO runs were rejected because they improved isolated cases while reducing overall response quality.

Codebase

https://github.com/EddyEjembi/Vital-Med-Tiny

Safety And Limitations

Vital-Med Tiny is for educational guidance only. It may hallucinate, omit important red flags, or give incomplete advice. It should not replace clinicians, emergency care, or national medical guidelines.

The model was designed to avoid definitive diagnosis and to recommend professional care when symptoms may be serious. Users should seek qualified medical care for severe, persistent, or worsening symptoms.

Citation

If you discuss this model, cite it as:

Vital-Med Tiny, Africa Deep Tech Challenge 2026 Laptop LLM submission by Eddy Ejembi.

Downloads last month: 13

GGUF

Model size

2B params

Architecture

qwen35

Hardware compatibility

4-bit