HyperQuest

HyperQuest

This project mimics HyperQuest from Stanford University, CS231n.

HyperQuest is a web-app used by beginners in Machine Learning to easily get a proper intuition for choosing the right hyperparameters. This is initially an extremely daunting task because not having proper hyperparameters leads to the models breaking down.

---

TODO

• Generate data
• Construct website with javascript
• Transfer all json files online
• Merge all javascript to one file
• Minimize the javascript
• Push the first beautiful version to github pages
• Add PV/UV to web
• Use brighter color
• Try flat design
• generate wechat share <meta>
• generate tweeter share meta
• Add more description to web
• Provide the easy-to-use model for other web pages
• Add 404 page
• Generate new model

---

Embed HyperQuest to your site

1. Use the anchor HTML:

   <div id = 'form'></div>
   <a type="button" href="javascript: submitForm()">Run</a>
   <span id="acc"></span>
   <p id="gradient_canvas"></p>
   <div id="loss_chart" style="height: 300px; width: 100%;"></div>
   <div id="train_and_vali_acc" style="height: 300px; width: 100%;"></div>

2. Use the HyperQuest Javascript (below the anchor HTML)

   1. Embed the 2-layer net

   <script type="text/javascript" src='https://cdn.plot.ly/plotly-latest.min.js'></script>
   <script type="text/javascript" src='https://bingcheng.openmc.cn/HyperQuest/html/2-layer-latest.js'></script>

   2. Embed the neural net

   <script type="text/javascript" src='https://cdn.plot.ly/plotly-latest.min.js'></script>
   <script type="text/javascript" src='https://bingcheng.openmc.cn/HyperQuest/convhtml/conv-latest.js'></script>

---

1. Two Layer Net

Try the simple version here.

2. Conv Neural Net

Try the convolutional version here.

Tuning Parameters is time-consuming?

Don't understand the usage of each parameter?

The meaning of the curve is difficult to understand?

HyperQuest can turn you into an expert!