AI Demo Template
Finished
Machine Learning
2024-08-30
A comprehensive template for building AI demonstrations with model training, interactive UI, and web deployment. Features a neural network trained on 1,000 examples over 10,000 epochs, achieving 97.54% loss reduction (final loss: 0.000460).
How to Use
Enter two numbers between -1 and 1 in the input fields and click 'Predict' to see the trained model's output. The model learns the function (input1 + input2).tanh() * 0.5 + 0.5, mapping the sum to [0,1] range.
Technical Details
Model Size: ~379 bytes (332 bytes weights + 47 bytes metadata)
Requirements: WebAssembly support, modern browser, 100MB RAM
Performance: Loads instantly with embedded model assets, responsive UI, optimized WASM build. Model trained in 13.3s with 97.54% loss reduction.
Technology Stack
Bevy
Candle
WASM
Rust
Keywords
ai
machine-learning
template
neural-network
bevy
candle
wasm
training