Lumi-node/infinite-context

Infinite Context - Live Demo

Give any LLM unlimited memory with sub-millisecond retrieval.

What This Demo Shows

This is a live demonstration of HAT (Hierarchical Attention Tree) - a retrieval system that:

• —100% accuracy finding relevant conversations
• —< 1ms search time across hundreds of thousands of tokens
• —1,400x context extension for small models

How to Use

1. 1.Click Initialize to create a simulated conversation history
2. 2.Ask natural questions like:
3. 3."What did we do to fix the React error?"
4. 4."How much did we speed up the Python script?"
5. 5."What was causing the Kubernetes pods to crash?"
6. 6.See HAT retrieve the exact relevant conversations in milliseconds

Performance

Links

• —GitHub
• —Docker Hub
• —ArXiv Paper (coming soon)

License

MIT