TinyTorch/demos/README.md

TinyTorch Demo System

This directory contains progressive AI capability demonstrations for TinyTorch. Each demo showcases what becomes possible as you export more modules to the TinyTorch package.

🎯 Available Demos

Run any demo using: tito demo <demo_name>

Core Demos

Demo	Command	Module Requirements	Description
Mathematical Operations	tito demo math	Module 02 (Tensor)	Linear algebra, matrix operations, geometric transformations
Logical Reasoning	tito demo logic	Module 03 (Activations)	Boolean functions, XOR problem, decision boundaries
Single Neuron Learning	tito demo neuron	Module 04 (Layers)	Watch a neuron learn the AND gate with gradient descent
Multi-Layer Networks	tito demo network	Module 05 (Dense)	Solve the famous XOR problem with 2-layer network
Computer Vision	tito demo vision	Module 06 (Spatial)	Image processing, edge detection, CNN pattern recognition
Attention Mechanisms	tito demo attention	Module 07 (Attention)	Sequence processing, self-attention, transformer foundations
End-to-End Training	tito demo training	Module 11 (Training)	Complete ML pipeline with optimization and evaluation
Language Generation	tito demo language	Module 16 (TinyGPT)	AI text generation and language modeling

Demo Commands

# Show capability matrix
tito demo

# Run specific demo
tito demo math
tito demo vision
tito demo attention

# Run all available demos
tito demo --all

# Show matrix only (no module testing)
tito demo --matrix

🚀 Demo Progression

The demos unlock progressively as you export modules:

Foundation (Modules 2-5)

• Tensor Math: Matrix operations, linear systems
• Activations: Nonlinear functions, sigmoid/ReLU
• Single Neuron: Gradient descent learning
• XOR Network: Multi-layer breakthrough

Intelligence (Modules 6-7)

• Computer Vision: CNNs, edge detection, pattern recognition
• Attention: Sequence understanding, transformer mechanisms

Complete Systems (Modules 11-16)

• Training: End-to-end ML pipelines
• Language: Text generation, TinyGPT

🎓 Educational Value

Each demo is designed to:

1. Show Real AI Capabilities: Not just code, but actual intelligence in action
2. Explain the "Why": Understanding principles behind the implementations
3. Connect to Production: How these concepts scale to real ML systems
4. Build Excitement: See your framework grow more capable with each module

🔧 Technical Details

• Import Safety: Each demo gracefully handles missing modules
• Error Recovery: Clear messages about which modules need to be exported
• Rich Output: Color-coded, formatted demonstrations with explanations
• Self-Contained: Each demo can run independently for testing

🌟 Demo Highlights

Mathematical Operations (demo_tensor_math.py)

• Solves real linear algebra problems
• Geometric transformations and rotations
• Preview of neural network computations

XOR Network (demo_xor_network.py)

• The classic AI milestone problem
• Shows why single layers fail
• Demonstrates hidden layer feature creation

Computer Vision (demo_vision.py)

• Edge detection with Sobel operators
• Convolutional pattern recognition
• Complete CNN architectures

Attention Mechanisms (demo_attention.py)

• Self-attention matrix computation
• Multi-head attention concepts
• Connection to modern language models

Language Generation (demo_language.py)

• Token embeddings and sequence processing
• Autoregressive generation process
• Complete transformer architecture overview

📈 Usage Analytics

The demo system tracks:

• Which modules are exported and available
• Demo availability status (✅ Ready, ⚡ Partial, ❌ Not Available)
• Integration with TinyTorch package exports

Students can see their progress through the capability matrix and immediately test new functionality as they complete modules.