metadata

title: Gemma AI Assistant
emoji: 🤖
colorFrom: indigo
colorTo: purple
sdk: docker
sdk_version: '1.0'
app_file: app.py
pinned: false

Gemma AI Assistant Space

This Space hosts the backend API for the Gemma AI Assistant, a conversational AI that combines local LLM processing using HuggingFace Transformers and real-time chat capabilities with Google's Gemini API.

Features

FastAPI backend with async support
Local LLM using mradermacher/Huihui-gemma-3n-E4B-it-abliterated-GGUF
Gemini API integration for real-time chat
Supabase integration for data persistence
Containerized deployment

API Endpoints

POST /api/chat

Process chat messages using either the local LLM or Gemini API.

Request Body:

{
  "messages": [
    {
      "role": "user",
      "content": "Hello, how are you?"
    }
  ],
  "use_gemini": true,
  "temperature": 0.7
}

Response:

{
  "response": "I'm doing well, thank you! How can I help you today?"
}

Environment Variables Required

GOOGLE_AI_STUDIO_KEY: Your Google AI Studio API key
SUPABASE_URL: Your Supabase project URL
SUPABASE_SERVICE_KEY: Your Supabase service role key
HF_MODEL_ID: HuggingFace model ID (default: mradermacher/Huihui-gemma-3n-E4B-it-abliterated-GGUF)

Local Development

Install dependencies:

pip install -r requirements.txt

Run the server:

uvicorn app:app --reload --port 7860

Testing

Run the tests using pytest:

pytest test_app.py -v