README.md

metadata

title: Ollama API Space
emoji: 🚀
colorFrom: blue
colorTo: purple
sdk: docker
app_port: 7860

🚀 Ollama API Space

A Hugging Face Space that provides a REST API interface for Ollama models, allowing you to run local LLMs through a web API.

🌟 Features

• Model Management: List and pull Ollama models
• Text Generation: Generate text using any available Ollama model
• REST API: Simple HTTP endpoints for easy integration
• Health Monitoring: Built-in health checks and status monitoring
• OpenWebUI Integration: Compatible with OpenWebUI for a full chat interface

🚀 Quick Start

1. Deploy to Hugging Face Spaces

1. Fork this repository or create a new Space
2. Upload these files to your Space
3. No environment variables needed - Ollama runs inside the Space!
4. Wait for the build to complete (may take 10-15 minutes due to Ollama installation)

2. Local Development

# Clone the repository
git clone <your-repo-url>
cd ollama-space

# Install dependencies
pip install -r requirements.txt

# Install Ollama locally
curl -fsSL https://ollama.ai/install.sh | sh

# Start Ollama in another terminal
ollama serve

# Run the application
python app.py

📡 API Endpoints

GET /api/models

List all available Ollama models.

Response:

{
  "status": "success",
  "models": ["llama2", "codellama", "neural-chat"],
  "count": 3
}

POST /api/models/pull

Pull a model from Ollama.

Request Body:

{
  "name": "llama2"
}

Response:

{
  "status": "success",
  "model": "llama2"
}

POST /api/generate

Generate text using a model.

Request Body:

{
  "model": "llama2",
  "prompt": "Hello, how are you?",
  "temperature": 0.7,
  "max_tokens": 100
}

Response:

{
  "status": "success",
  "response": "Hello! I'm doing well, thank you for asking...",
  "model": "llama2",
  "usage": {
    "prompt_tokens": 7,
    "completion_tokens": 15,
    "total_tokens": 22
  }
}

GET /health

Health check endpoint.

Response:

{
  "status": "healthy",
  "ollama_connection": "connected",
  "available_models": 3
}

🔧 Configuration

Environment Variables

• OLLAMA_BASE_URL: URL to your Ollama instance (default: http://localhost:11434 - Ollama runs inside this Space!)
• MODELS_DIR: Directory for storing models (default: /models)
• ALLOWED_MODELS: Comma-separated list of allowed models (default: all models)

Note: This Space now includes Ollama installed directly inside it, so you don't need an external Ollama instance!

Supported Models

By default, the following models are allowed:

• llama2
• llama2:13b
• llama2:70b
• codellama
• neural-chat

You can customize this list by setting the ALLOWED_MODELS environment variable.

🌐 Integration with OpenWebUI

This Space is designed to work seamlessly with OpenWebUI. You can:

1. Use this Space as a backend API for OpenWebUI
2. Configure OpenWebUI to connect to this Space's endpoints
3. Enjoy a full chat interface with your local Ollama models

🐳 Docker Support

The Space includes a Dockerfile for containerized deployment:

# Build the image
docker build -t ollama-space .

# Run the container
docker run -p 7860:7860 -e OLLAMA_BASE_URL=http://host.docker.internal:11434 ollama-space

🔒 Security Considerations

• The Space only allows access to models specified in ALLOWED_MODELS
• All API endpoints are publicly accessible (consider adding authentication for production use)
• The Space connects to your Ollama instance - ensure proper network security

🚨 Troubleshooting

Common Issues

1. Connection to Ollama failed: Check if Ollama is running and accessible
2. Model not found: Ensure the model is available in your Ollama instance
3. Timeout errors: Large models may take time to load - increase timeout values

Health Check

Use the /health endpoint to monitor the Space's status and Ollama connection.

📝 License

This project is open source and available under the MIT License.

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📞 Support

If you encounter any issues or have questions, please open an issue on the repository.