Back to Web Apps

AI Playground

Interactive AI experiments and machine learning demonstrations

Choose Your Experiment

Explore AI and machine learning concepts through hands-on interactive demonstrations. Each experiment runs entirely in your browser using WebLLM technology.

CNN Trainer
Train your own Convolutional Neural Network on real MNIST data. Watch your model learn to recognize handwritten digits with customizable training parameters.
Ready to Use
Neural Network Visualizer
Draw digits and watch a neural network classify them in real-time. See how artificial neurons process visual information.
Coming Soon
Computer Vision Lab
Upload images or use your webcam for real-time object detection, edge detection, and visual analysis.
Coming Soon
NLP Playground
Analyze text sentiment, generate content, detect languages, and explore natural language processing.
Coming Soon
Algorithm Visualizer
Watch sorting algorithms, pathfinding, and search algorithms animate step-by-step.
Coming Soon
Data Science Studio
Explore clustering, regression, pattern recognition, and interactive data analysis.
Coming Soon
AI Model Comparison
Compare different AI models, benchmark performance, and explore bias detection.
Coming Soon

Neural Network Trainer

Back to Experiments

Train Your Own CNN

Train a Convolutional Neural Network on real MNIST handwritten digit data. Watch as your model learns to recognize digits!

Training Configuration

What You're Building

You're training a real Convolutional Neural Network (CNN) on the famous MNIST dataset - the same data used to benchmark machine learning models worldwide.

Your CNN will learn to recognize handwritten digits by finding patterns in pixel data through multiple layers of artificial neurons.

CNN Architecture

Input: 28×28 grayscale images

Conv2D Layer 1: 16 filters, 3×3 kernel

MaxPool2D: 2×2 pooling

Conv2D Layer 2: 32 filters, 3×3 kernel

MaxPool2D: 2×2 pooling

Dense: 128 hidden units

Output: 10 classes (digits 0-9)

Training Data

Ready to download MNIST data...

Model Performance

Train your model to see performance metrics!