Neural Network Builder in WebAssembly

Loading the Neural Network Builder library. Please wait ...

To run this demo, use a browser which has support for WebAssembly, and optionally SIMD if you checked that option.

To use WebAssembly SIMD instructions on Google Chrome, open the application from the terminal and pass to it --js-flags="--experimental-wasm-simd"

Our experiments were done using Google Chrome 73.0.3683.86

Open your browser console to see the model results.

Layers code

A - Model configuration

Training batch

Testing batch

Prediction batch

Loss function

Training batch in memory

Testing batch in memory

B - Training parameters

Epoch

Learning rate

L1 Regularizer

L2 Regularizer

Training data

Training label

C - Testing parameters

Testing data

Testing label

Use SIMD Log forward Log backward

Prediction data