Google Gemma for Web

A web-based implementation of Google's Gemma language model using MediaPipe Tasks for GenAI.

Overview

This project demonstrates how to run Google's Gemma language model directly in the browser using WebAssembly. It provides a simple interface for text generation using the Gemma 2B model.

Features

• Browser-based inference using WebAssembly
• No server required - runs completely client-side
• Simple text input interface
• Real-time text generation

Prerequisites

• Modern web browser with WebAssembly support
• Kaggle account (for model download)

Setup Instructions

1. Clone the repository:

   git clone https://github.com/atomostechhq/google-gemma-for-web-exp.git
   cd google-gemma-for-web-exp

2. Download the model:

   • Follow the instructions in assets/instructions.txt to:
     • Set up Kaggle authentication
     • Download the model file
     • Place it in the correct location

3. Serve the application:

   • Use any HTTP server to serve the files locally
   • For example, with Python:

     python3 -m http.server 8000

   • Or with Node.js:

     npx http-server

4. Access the application:

   • Open your browser and navigate to http://localhost:8000

Usage

1. Enter your text prompt in the input field
2. Click "Get Response" to generate text
3. The model will stream its response in real-time

Technical Details

• Uses MediaPipe Tasks for GenAI
• Implements the Gemma 2B model quantized to INT8
• Runs inference directly in the browser using WebAssembly

Limitations

• Initial model loading may take some time depending on your internet connection
• Performance depends on your device's capabilities

License

This project uses Google's Gemma model which is subject to its own license terms. Please refer to the Gemma documentation for details.