🧠 CUP-Framework

Universal Parametric Brain Architecture (CUP / CUP++ / CUP++++)

CUP-Framework is a system of compact, invertible, and modular neural brains, designed for researchers, students, and enthusiasts in symbolic, logical, or mathematical artificial intelligence.

🔍 Overview

This project is based on precise mathematical formulas allowing:

  • Learning or reproducing symbolic, logical, or analytical functions

  • Reversing outputs to reconstruct the original inputs

  • Stacking multiple brains into reasoning chains

  • Saving, loading, and sharing each brain as a standalone module

🧠 Three Levels of Architecture

  • CUP: minimal and analytically invertible core, 2 tanh layers, fast and lightweight

  • CUP++: adds contextual modulation masks (M₁, M₂) for adaptive inputs

  • CUP++++: adds layer normalization, residual connections, and parametric activations (e.g. tanh, sigmoid, sinh)

🧪 Use Cases

  • Logical or arithmetic neural calculators (counting, multiplying, dividing)

  • Reconstructing symbolic inputs from a known result

  • Modular brain chaining (e.g., CUP++ → CUP++ → CUP++ with inversion support)

  • Embedding compact brains in simulators or autonomous AI agents

📦 Download

👉 Current Version:
🔗 CUP-Framework v1.0.0 on GitHub

This release includes:

  • CUP neural core compilable as .pyd / .so

  • Cython source code (.pyx)

  • A complete Python usage example (test_cup.py)

  • Clear README.md with usage instructions

🔐 License & Usage

  • ✅ Free for scientific, educational, and student use

  • 🚫 Commercial or industrial use is strictly prohibited without explicit license

For business inquiries or licensing, please contact the project author directly.

Scroll to Top
Skip to content