Skip to main content

Welcome to EdgeAI

Welcome to EdgeAI Documentation - the first Android app to demonstrate real Llama3.2-1B inference using ExecuTorch with Qualcomm QNN backend.

What is EdgeAI?​

EdgeAI is a groundbreaking Android application that brings real AI inference to mobile devices. Unlike simulated or rule-based approaches, EdgeAI runs actual Llama3.2-1B model inference on your Android device using:

  • ExecuTorch: PyTorch's mobile-optimized runtime
  • Qualcomm QNN: Hardware acceleration via Qualcomm AI Engine Direct
  • v79 Context Binaries: Optimized for SoC Model-69
  • ARM64-v8a: Native Android architecture support

Key Features​

✅ Real Model Inference - Actual Llama3.2-1B running on device
✅ Hardware Acceleration - Qualcomm QNN backend for optimal performance
✅ Professional Documentation - Comprehensive guides and technical details
✅ Open Source - Full source code available on GitHub
✅ Mobile Optimized - Designed specifically for Android devices

Quick Start​

Get started with EdgeAI in just a few steps:

  1. Setup Qualcomm AI HUB - Configure your development environment
  2. Understand Project Structure - Learn how EdgeAI is organized
  3. Check Latest Release - See what's new in the latest version

Architecture Overview​

EdgeAI uses a sophisticated architecture combining:

  • Android App Layer: Kotlin/Java UI and application logic
  • Native C++ Layer: High-performance inference engine
  • ExecuTorch Runtime: PyTorch mobile optimization
  • Qualcomm QNN Backend: Hardware acceleration
  • Model Layer: Llama3.2-1B with optimized context binaries

Getting Help​

  • 📖 Documentation: Comprehensive guides and technical details
  • 🐛 Issues: Report bugs and request features on GitHub
  • 💬 Discussions: Join the community discussions
  • 📧 Contact: Reach out for support and collaboration

What's Next?​

Ready to dive deeper? Check out our technical documentation:

Built with ❤️ for the AI community