ACLAS Neuro-Edu SDK

Autonomous Cognitive Simulation Framework for AI-Powered Education

by Atlanta College of Liberal Arts and Sciences (ACLAS)

---

English | Français | Español | Русский | Deutsch | 繁體中文 | 日本語 | 한국어

---

---

🌐 Website  |  🚀 Live Demo  |  🤗 Hugging Face  |  📚 Courses  |  🏆 Certifications  |  💬 Community  |  📄 Docs

---

🌌 What is ACLAS Neuro-Edu?

ACLAS Neuro-Edu is a research-grade, open-source multi-agent cognitive simulation framework developed by Atlanta College of Liberal Arts and Sciences. It models the process of human knowledge acquisition using on-device neural networks, vector-space semantics, and thermodynamic entropy analysis.

Now enhanced with Hugging Face Hub integration, the framework supports exporting cognitive model weights for collaborative research and federated evaluation.

📐 Mathematical Foundation

Neuro-Edu operationalizes cognitive science principles into computable alignment objectives.

1. Intrinsic Load Estimation (CLT)

To manage working memory capacity, we estimate the intrinsic load $L_{intrinsic}$ of instructional content: $$L_{intrinsic} = \sum_{i=1}^{n} w_i \cdot F_i(x) + \epsilon$$ where $F_i(x)$ represents lexical complexity, syntactic depth, and conceptual density.

2. Cognitive-Grounded Alignment Protocol (CGAP)

We extend the standard RLHF objective to penalize cognitive misalignment: $$\mathcal{L}{CGAP} = \mathbb{E}[r(x,y)] - \beta D{KL}(\pi || \pi_{ref}) - \lambda_1 L_{load} - \lambda_2 L_{metacog}$$ This ensures that the AI tutor optimizes for learning effectiveness rather than just perceived helpfulness.

Why does this exist?

At ACLAS, we believe the future of education is deeply intertwined with AI. This framework is our commitment to that belief — a live, open, runnable proof that AI and pedagogy can be unified at the systems level.

---

✨ Key Features

Feature	Description
🧠 On-Device Neural Kernel	Pure NumPy MLP with real backpropagation (He init, ReLU, Sigmoid, SGD)
🤗 Hugging Face Hub	Official model repository for sharing and versioning cognitive weights
📡 Multi-Agent Social Learning	20+ autonomous agents exchange knowledge via a real-time message bus
📊 Cognitive Entropy Analytics	Thermodynamic-inspired classroom entropy tracking (Shannon H, MSE loss)
🛡️ Industrial-Grade CI/CD	Automated security (Bandit), dependency audit (pip-audit), and linting (Ruff)
📈 Performance Benchmarks	Real-time tracking of tick latency (500+ ticks/s) and training efficiency
🎯 5-Dim Skill Matrix	Logic / Math / Language / Memory / Creative skill profiles
🕸️ Knowledge Graph	D3.js-powered live concept relation map built from agent learning events
✅ Regression Testing	Comprehensive suite covering Neural Kernels, Social Bus, and FastAPI Endpoints

---

🔌 Pluggable LLM Cognitive Engines

While Neuro-Edu ships natively with a minimalist, first-principles neural network (TinyCognitionModel), our framework is designed to be highly extensible. You can seamlessly plug in Ollama or vLLM to power each AI student's internal monologue and thought process using the world's leading open-source models!

We natively support and highly recommend testing the following open-weights models as cognitive engines within our multi-agent sandbox:

• 🦙 Llama-3 & Llama-3.1 (Meta)
• 🤖 Qwen 2.5 (Alibaba Cloud) - Default configured in our agent.py
• 🐋 DeepSeek-V3 / DeepSeek-Coder (DeepSeek)
• 💎 Gemma-2 (Google)
• 🔬 Phi-3 (Microsoft)
• 🌪️ Mistral / Mixtral (Mistral AI)

Build your own LLM arena and see how different models react to educational stimuli!

---

🏗️ Architecture

graph TD
    %% Global Styles
    classDef layer fill:#f9f9f9,stroke:#333,stroke-width:2px,color:#000;
    classDef core fill:#e1f5fe,stroke:#01579b,stroke-width:2px,color:#01579b,font-weight:bold;
    classDef frontend fill:#fff3e0,stroke:#e65100,stroke-width:2px,color:#e65100;
    classDef api fill:#f3e5f5,stroke:#4a148c,stroke-width:2px,color:#4a148c;

    subgraph SDK ["ACLAS Neuro-Edu Framework Architecture"]
        direction TB

        subgraph L0 ["Layer 0: Visualization"]
            direction LR
            V1["WebGL / 3D Nebula"] --- V2["Three.js Frontend"]
        end

        subgraph L1 ["Layer 1: Neural Kernel"]
            direction LR
            N1["NumPy MLP"] --- N2["Backpropagation Engine"]
        end

        subgraph L2 ["Layer 2: Cognitive Profiles"]
            direction LR
            S1["5-Dim Skill Matrix"] --- S2["Agent DNA"]
        end

        subgraph L3 ["Layer 3: Core Engine"]
            direction TB
            E1["UltimateClassroom Bus"]
            E2["Social Learning Bus"]
            E3["Federated Training"]
            E4["Entropy & GPA Evaluator"]
            E1 --- E2 --- E3 --- E4
        end

        subgraph L4 ["Layer 4: API Gateway"]
            A1["FastAPI REST Layer"]
            A2["/api/teach · /api/train · /api/metrics"]
            A1 --- A2
        end

        %% Connections
        L2 --> L1
        L1 --> L0
        L3 --> L2
        L3 --> L1
        L4 --> L3
    end

    %% Apply Classes
    class L0,L1,L2 layer;
    class L3 core;
    class L4 api;
    class SDK frontend;

Loading

File Structure

aclas-neuro-edu/
├── neuro_edu/             # Source code (PyPI namespace)
│   ├── model.py          # TinyCognitionModel — NumPy MLP w/ real backprop
│   ├── agent.py          # NeuralStudentAgent — LLM thought + skill matrix
│   ├── engine.py         # UltimateClassroom — social learning bus + metrics
│   ├── evaluator.py      # GPA, CAS, Retention, Shannon Diversity, Dropout Risk
│   ├── skills.py         # 5-dim SkillMatrix with 6 agent profiles
│   └── knowledge_graph.py # Lightweight concept graph for D3.js
├── data/
│   └── knowledge_base.json  # 50-sample labeled training dataset
├── web/
│   ├── index.html        # Full dashboard: 3D + Charts + Telemetry
│   └── app.js            # Three.js · Chart.js · D3.js frontend logic
├── tests/
│   ├── test_model.py     # 9 tests — backprop correctness, loss descent
│   ├── test_agent.py     # 7 tests — initialization, mood, memory
│   └── test_engine.py    # 10 tests — classroom, training, metrics
├── cli.py                # Command-line interface
├── main.py               # FastAPI application entry point
└── requirements.txt

---

🚀 Quick Start

Prerequisites

• Python 3.10+
• pip

Installation

git clone https://github.com/aclascollege/neuro-edu.git
cd neuro-edu
pip install -r requirements.txt

Launch the Dashboard

python main.py

Open http://localhost:8000 in your browser.

---

🖥️ CLI Usage

# Simulate a classroom broadcast
python cli.py simulate --text "Introduction to Bayesian inference" --complexity 0.7 --domain math --agents 20

# Run federated training on the dataset
python cli.py train --epochs 10

# Print full evaluation report
python cli.py report

Example training output:

  Epoch  Loss
  ──────────────────────
      1  0.182341  ██████████████████████
      2  0.143892  ██████████████████
      3  0.118776  ███████████████
      4  0.099234  ████████████
      5  0.084102  ██████████

  Final loss: 0.084102
  Training complete.

---

📡 REST API Reference

Method	Endpoint	Description
GET	/api/status	Live state of all agents
POST	/api/teach	Broadcast instruction to all agents
POST	/api/train	Run federated training (real backprop)
GET	/api/metrics	Full evaluation report (GPA, CAS, etc.)
GET	/api/graph	D3.js knowledge graph export
GET	/api/architecture	Neural network architecture summary
POST	/api/reset	Reset simulation environment

Example: Teach

curl -X POST http://localhost:8000/api/teach \
  -H "Content-Type: application/json" \
  -d '{"text": "Quantum entanglement fundamentals", "complexity": 0.85, "skills_req": {"math": 0.9}}'

Example: Train

curl -X POST http://localhost:8000/api/train
# Returns: {"epochs": 5, "epoch_losses": [0.18, 0.14, 0.11, 0.09, 0.08], "final_loss": 0.08}

---

🧪 Testing

pytest tests/ -v

tests/test_model.py::TestNeuralLayer::test_forward_shape              PASSED
tests/test_model.py::TestTinyCognitionModel::test_loss_decreases      PASSED
tests/test_agent.py::TestNeuralStudentAgent::test_initialization      PASSED
tests/test_engine.py::TestUltimateClassroom::test_training            PASSED
...
26 passed in 0.52s

---

📐 The Science Behind It

Cognitive Absorption Model

Each agent predicts its absorption probability using a 3-layer MLP:

Input: [complexity, attention, skill_match, fatigue, prior_knowledge]
  ↓  Dense(16) + ReLU
  ↓  Dense(8) + ReLU
  ↓  Dense(1) + Sigmoid
Output: absorption_probability ∈ (0, 1)

Evaluation Metrics

Metric	Formula
GPA	(knowledge_depth × 0.5 + attention × 0.3 + prior × 0.2) × 4.0
CAS Score	Σ(absorption_i × attention_i) / n
Retention Rate	count(agents with knowledge_pool > 0 AND mood ≠ Confused) / n
Diversity Index	Shannon entropy H = -Σ p(mood) × log₂(p(mood))
Dropout Risk	(1−attention)×0.5 + fatigue×0.3 + (1−prior)×0.2

Social Learning Bus

When an agent reaches mood = Excited (absorption > 68%), it broadcasts a clarity signal to up to 2 confused peers. This models peer-to-peer knowledge diffusion observed in real classrooms.

---

🤝 Contributing

We welcome contributions from researchers, developers, and educators!

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Please read our Contributing Guidelines and ensure all tests pass before submitting.

Ideas for Contributions

• 🔌 Ollama / local LLM integration for real agent thoughts
• 🗄️ ChromaDB / pgvector persistent knowledge store
• 🤖 RL-based teacher agent (learns optimal instruction complexity)
• 🌍 Multi-language support for the dashboard
• 📱 Mobile-responsive UI

---

🌍 Our Mission, Vision & The Open-Source Narrative

"Every mind, regardless of origin or means, deserves world-class learning." — ACLAS Mission & Vision

Making Elite Education a Universal Human Right

Atlanta College of Liberal Arts and Sciences (ACLAS) was founded on a radical conviction: elite education should be a universal human right, not a privilege gated by geography or economic status.

We are the world's first antifragile university — an online institution unbound by geographic limits, ensuring equal opportunities for all, from San Francisco to South Sudan, with one unwavering promise:

Pillar	What We Do
🌐 Global Equity Pricing	Fixed global tuition, 100% transparent, flexible installments
🤖 AI-Powered Classroom	150+ nation learning grid connecting conflict zones to megacities
⚖️ Geopolitical Neutrality	Accessible in all territories via decentralized tech, 24/7 Digital Embassy
🔗 Blockchain Credentials	Degrees with blockchain-tracked social impact metrics, fraud-proof

Our ACLAS Vision Statement:

ACLAS is the world's first antifragile education — an online university unbound by geographic limits, ensuring equal opportunities for all. We deliver Ivy League rigor through radical equity.
Neural Network Curriculum: Self-evolving online framework adapting to regional needs and aligned with UN Sustainable Development Goals.

---

🧬 Why We Build Open-Source AI

ACLAS Neuro-Edu isn't just a demo. It is a statement.

At a time when AI in education means paying for an API wrapper, we chose a different path: build the math from scratch, publish the code, and let anyone inspect every gradient descent step.

This project is the technical embodiment of our institutional mission:

• Radical Transparency — Every weight update is observable. No black-box inference. You can read the backpropagation yourself.
• Access Over Gatekeeping — By open-sourcing this framework under MIT License, we ensure any researcher, student, or engineer anywhere in the world can run, fork, and extend our cognitive simulation tools.
• AI for Human Flourishing — We don't build AI to replace teachers. We model cognition to understand how knowledge transfers between human minds — and to make that process better.
• Credibility Through Code — The best argument for ACLAS's academic seriousness is not a brochure. It's a codebase with 26 passing tests, real backpropagation, and a live WebGL visualization of multi-agent learning.

In 2026, as AI reshapes every layer of society, ACLAS is planting its flag: education institutions must lead AI development, not merely consume it.

"We don't wrap GPT. We model cognition from first principles."

---

🎓 About ACLAS

Atlanta College of Liberal Arts and Sciences (ACLAS) is a forward-thinking institution at the intersection of liberal arts and cutting-edge technology. We provide globally accessible, accredited education with a deep commitment to AI-powered learning innovation.

🔗 Explore ACLAS

Resource	Link
🌐 Official Website	aclas.college
📚 Online Courses	aclas.college/courses
🏆 Certifications	aclas.college/certification
🎓 Academic Programs	aclas.college/programs
📰 Research & News	aclas.college/news
💼 Careers	aclas.college/careers
📩 Contact	aclas.college/contact

📱 Follow ACLAS

---

📄 Citation

If you use this framework in your research, please cite:

@software{aclas_neuro_edu_2026,
  title     = {ACLAS Neuro-Edu: Autonomous Cognitive Simulation Framework},
  author    = {{Atlanta College of Liberal Arts and Sciences}},
  year      = {2026},
  url       = {https://github.com/aclascollege/neuro-edu},
  note      = {Open-source AI education research framework},
  institution = {ACLAS, \url{https://aclas.college}}
}

---

📜 License

This project is licensed under the MIT License — see the LICENSE file for details.

You are free to use, modify, and distribute this software.

🌍 Community & Support

We are building the first Anti-Fragile University together.

🏆 Awesome Lists Support

Help us increase visibility by submitting Neuro-Edu to curated lists! We are already preparing submissions for:

• awesome-machine-learning
• awesome-neuroscience
• awesome-education

See our Contribution Guide for details on how to help.

🏛️ About ACLAS College

The Atlanta College of Liberal Arts and Sciences (ACLAS) is a leading institution dedicated to bridging the gap between classical liberal arts and cutting-edge artificial intelligence.

---

Built with ❤️ by Atlanta College of Liberal Arts and Sciences

Advancing human potential through AI-powered education

🌐 aclas.college  ·  📚 Courses  ·  🏆 Certifications  ·  🌍 Mission & Vision  ·  💬 Contact

_{© 2026 Atlanta College of Liberal Arts and Sciences. Open-sourced under MIT License.  ·  Terms  ·  Privacy  ·  Sitemap}