CGAN_jittor

This repo comes from the competition of Jittor, where we put the codes of warm-up match.

Brief Introduction

This repo provides a model which is able to produce graph of numbers from random noise after proper training with the dataset MINIST.

As for an example, the picture below show the number which is generated by this model.

How to Install and Run

To run this model, you are able to clone the repo and run CGAN.py in competition/warm_up_comp.

Moreover, you are also able to get it from the source repo of jittor as below.

git clone https://github.com/Jittor/gan-jittor.git
cd gan-jittor/
sudo python3.7 -m pip install -r requirements.txt
cd competition/warm_up_comp
修改 CGAN.py 使其运行

Structure

The descriptions about each directory are as the following:

–CGAN_jittor (root)

1. -asset: graphs which used for the trainning
2. -data: scripts which used to download the dataset
3. -model: model which will be trained
4. -competition: other part of the model