Developer: Bojan Milanović (@bis3946 - Root Authority)
Core Framework: NuN Nexus v4.9 Core
TTF-9 is an autonomous, post-quantum resistant data integrity auditor designed for extreme hardware efficiency and zero-entropy data filtration. Abandoning traditional 32-bit floating-point networks, it utilizes a 1.58-bit ternary quantization-inspired model governed by Triadic Logic.
The engine actively audits, repairs, and archives unstructured data (PDFs, TXT, CSV) using an Active Inference loop, functioning as a fully autonomous digital cortex.
Every process within the TTF-9 architecture is governed by the Triadic Equilibrium formula:
F(x,y,z) = (x * y * z)^(1/3)
- x (Generation): Logical possibility and coherence.
- y (Stability): Factual reality and empirical grounding.
- z (Equilibrium): Absence of subjective bias or malicious injection.
Variables are strictly bound to {1, 0, -1}. A result of F=1 grants system approval. Any entropy (0) or hostile logic (-1) triggers the Repair Engine.
- Universal Auditor: Processes PDFs, TXT, and CSV files natively.
- OCR Vision: Integrates Tesseract OCR and Poppler to read rasterized/scanned PDFs.
- Active Inference (Self-Correction): The system does not merely reject false data; it autonomously rewrites and repairs semantic errors to achieve Triadic Equilibrium.
- Persistent Knowledge Base: Integrates directly with Google Drive to cache learned repairs (
MEMORY_HIT), radically reducing API token consumption over time. - Clean Library Export: Outputs fully verified and repaired continuous TXT documents and detailed CSV audit logs.
- Open the
TTF9_Universal_Auditor_Public.ipynbin Google Colab. - Run the cell to install system dependencies (Tesseract, Poppler).
- The script will request Google Drive access to construct the
TTF9_Systemfolder hierarchy. - Input your Groq API Key when prompted securely.
- Upload your target document and let the Active Inference loop process the data.
All verified data is permanently securely stored in your Google Drive under:
/MyDrive/TTF9_System/Universal_Clean_Library/