README.md

metadata

title: CASCADE Hyperlattice
emoji: 🌌
colorFrom: purple
colorTo: blue
sdk: docker
app_port: 8501
thumbnail: thumbnail.png
tags:
  - streamlit
  - cascade-lattice
  - quine
  - 3d-visualization
  - rerun
  - glass-box-ai
pinned: true
short_description: Real-time quine swarm visualization with 4D provenance

🌌 CASCADE Hyperlattice Explorer

🔮 GLASS BOX AI - Every weight, decision, and mutation is fully transparent

Real-time 3D visualization of quine agent swarms navigating a decision lattice with null gates (wormholes).

Features

Core Visualization

• 🧬 Quine Agents - Self-replicating agents that fork at null gates
• 🌀 Null Gates - Wormholes connecting distant lattice regions
• 🎨 5 Themes - Neon Vampire, Cyberpunk, Matrix, Solar Flare, Ice Palace
• ▶️ Real-time Animation - Smooth streaming visualization
• 📜 Event Log - Track forks, jumps, and pruning

CASCADE-LATTICE Integration

• 🔗 Merkle Provenance - Every decision is cryptographically hashed
• 📊 Causation Graphs - Track decision ancestry and lineage
• 👻 Ghost Tape - Complete event log with timestamps
• 🛑 HOLD Protocol - Human-in-the-loop oversight points

Rerun 4D Visualization (Optional)

• 👁️ Time Scrubbing - Navigate through decision history
• 🔍 Drillable Data - Click any agent/node for details
• 📈 Metrics Dashboards - Fitness, holds, overrides over time
• 🌲 Provenance Trees - Visual Merkle chain exploration

Champion Model

• 🏆 DreamerV3 RSSM - Real DeepMind architecture (5120-dim latent)
• 🧠 LoRA Adapter - ~82K evolved parameters
• ✅ Quine Verified - Merkle hash: 38438eb04586975cda66a08fd3c447bb

Quick Start

# Clone and run locally
docker build -t cascade-hyperlattice .
docker run -p 8501:8501 -p 9090:9090 cascade-hyperlattice

# With GPU support
docker build -f Dockerfile.gpu -t cascade-hyperlattice-gpu .
docker run --gpus all -p 8501:8501 -p 9090:9090 cascade-hyperlattice-gpu

Ports

Port	Service
8501	Streamlit main app
9090	Rerun web viewer

---

📦 Model Repository

The champion quine brain is hosted at:

🔮 tostido/key-data/models — Download the model, run locally, verify integrity

from huggingface_hub import hf_hub_download

path = hf_hub_download(
    repo_id="tostido/key-data",
    filename="models/champion_gen42.py",
    repo_type="dataset"
)

---

💼 Quine Brain Conversion Service

Want YOUR model wrapped as a Glass Box Quine Brain?

• ✅ Self-contained single-file capsule
• ✅ Merkle-hashed provenance (tamper-proof)
• ✅ HOLD protocol (human override at inference)
• ✅ Full transparency APIs

📧 Contact for pricing and details