The hello-world-tensorspace from edmonton-school-of-ai

edmonton-school-of-ai / hello-world-tensorspace Goto Github PK

View Code? Open in Web Editor NEW

Played with Tensorspace a library for Neural network 3D visualization, building interactive and intuitive models in browsers, supports pre-trained deep learning models from TensorFlow, Keras, TensorFlow.js

JavaScript 99.92% HTML 0.08%

hello-world-tensorspace's Introduction

Hello Tensorspace

Click me to explore the model structure.

Resources

Check out the website to explore and play with other models.
Tensorspace - Github
Tensorspace - Website

Model Structure

The default view of the model structure is as shown below with corresponding labeled layers.

  const model = new TSP.models.Sequential(modelContainer);
 
  //Order corresponds with the image above default view
  model.add( new TSP.layers.GreyscaleInput({ shape: [28, 28, 1] }) );            //1
  model.add( new TSP.layers.Padding2d({ padding: [2, 2] }) );                    //2
  model.add( new TSP.layers.Conv2d({ kernelSize: 5, filters: 6, strides: 1 }) ); //3
  model.add( new TSP.layers.Pooling2d({ poolSize: [2, 2], strides: [2, 2] }) );  //4
  model.add( new TSP.layers.Conv2d({ kernelSize: 5, filters: 16, strides: 1 }) );//5
  model.add( new TSP.layers.Pooling2d({ poolSize: [2, 2], strides: [2, 2] }) );  //6
  model.add( new TSP.layers.Dense({ units: 120 }) );                             //7
  model.add( new TSP.layers.Dense({ units: 84 }) );                              //8
  model.add( new TSP.layers.Output1d({                                           //9
    units: 10, 
    outputs: ["0", "1", "2", "3", "4", "5", "6", "7", "8", "9"]
  }) );

Recommend Projects