# GAN Lab: An Interactive, Visual Experimentation Tool for Generative Adversarial Networks

By 
[Minsuk Kahng](http://minsuk.com),
[Nikhil Thorat](https://twitter.com/nsthorat),
[Polo Chau](https://www.cc.gatech.edu/~dchau/),
[Fernanda Viégas](http://fernandaviegas.com/), and 
[Martin Wattenberg](http://www.bewitched.com/)

## Overview

GAN Lab is a novel interactive visualization tool for anyone to learn and experiment with Generative Adversarial Networks (GANs), a popular class of complex deep learning models. With GAN Lab, you can interactively train GAN models for 2D data distributions and visualize their inner-workings, similar to [TensorFlow Playground](http://playground.tensorflow.org/).

GAN Lab uses [TensorFlow.js](https://js.tensorflow.org/), an in-browser GPU-accelerated deep learning library. Everything, from model training to visualization, is implemented with JavaScript. Users only need a web browser like Chrome to run GAN Lab. Our implementation approach significantly broadens people's access to interactive tools for deep learning. 

![Screenshot of GAN Lab](ganlab-teaser.png)


## Working Demo

Click the following link:

[https://poloclub.github.io/ganlab/](https://poloclub.github.io/ganlab/)

It runs on most modern web browsers. We suggest you use Google Chrome.


## Development

This section describes how you can develop GAN Lab.

### Install Dependencies

Run the following commands: 

```bash
$ git clone https://github.com/poloclub/ganlab.git
$ cd ganlab
$ yarn prep
```

It's unlikely, but you may need to install some basic JavaScript-related dependencies (e.g., yarn).


### Running Your Demo

Run the following command:

```bash
$ ./scripts/watch-demo

>> Waiting for initial compile...
>> 3462522 bytes written to demo/bundle.js (2.17 seconds) at 00:00:00
>> Starting up http-server, serving ./
>> Available on:
>>   http://127.0.0.1:8080
>> Hit CTRL-C to stop the server
```

Then visit `http://localhost:8080/demo/`. 

The `watch-demo` script monitors for changes of typescript code (e.g., `demo/ganlab.ts`)
and compiles the code for you.


## Credit

GAN Lab was created by 
[Minsuk Kahng](http://minsuk.com),
[Nikhil Thorat](https://twitter.com/nsthorat),
[Polo Chau](https://www.cc.gatech.edu/~dchau/),
[Fernanda Viégas](http://www.fernandaviegas.com/), and 
[Martin Wattenberg](http://www.bewitched.com/),
which was the result of a research collaboration between Georgia Tech and Google Brain/[PAIR](https://ai.google/research/teams/brain/pair).
We also thank Shan Carter and Daniel Smilkov, 
[Google Big Picture team](https://research.google.com/bigpicture/) and 
[Google People + AI Research (PAIR)](https://ai.google/research/teams/brain/pair), and 
[Georgia Tech Visualization Lab](http://vis.gatech.edu/)
for their feedback.
 
For more information, check out 
[our research paper](http://minsuk.com/research/papers/kahng-ganlab-vast2018.pdf):     
 
[Minsuk Kahng](http://minsuk.com),
[Nikhil Thorat](https://twitter.com/nsthorat),
[Polo Chau](https://www.cc.gatech.edu/~dchau/),
[Fernanda Viégas](http://www.fernandaviegas.com/), and 
[Martin Wattenberg](http://www.bewitched.com/).
"GAN Lab: Understanding Complex Deep Generative Models using Interactive Visual Experimentation."
*IEEE Transactions on Visualization and Computer Graphics, 25(1) ([VAST 2018](http://ieeevis.org/year/2018/welcome))*, Jan. 2019.


## 如何打包

## 注意事项

### 关于本地的Node环境
这个程序采用本地编译，然后在将dist复制到容器的方式，所以要先确保你的本地环境是
- node 8.17.0




### 测试运行
```
$ ./scripts/watch-demo
Waiting for initial compile...
3310042 bytes written to demo\bundle.js (2.25 seconds) at 16:51:11
Starting up http-server, serving ./
Available on:
  http://192.168.16.1:8080
  http://192.168.56.1:8080
  http://192.168.1.6:8080
  http://127.0.0.1:8080
Hit CTRL-C to stop the server
```
## 部署

```bash
$ ./scripts/deploy-demo
Stored bundle in output/bundle.js
Saved bundled demo at output/index.html
```

## 编译
```bash
$ ./scripts/build-demo
Stored bundle in output\bundle.js
```

## 制作容器镜像

```bash
 docker build -t ganlab .
```

## 运行容器
```bash
docker run -it  -p 8080:8080 ganlab
```

## 在容器重测试成功之后，可以导出，然后上传发布

```bash
docker save -o output/ganlab.tar ganlab:latest
```