A Jittor implementation of a Conditional GAN (CGAN)
This code is based on a framework obtained from www.educoder.net/competitions/index/Jittor-6 for 第三届计图人工智能挑战赛. It is the warmup question “计图挑战热身赛”.
The goal is to train a CGAN with the MNIST dataset, mapping noise and adding labels to number images from MNIST, that can then generate an image displaying the given input of numbers. The provided code framework (CGAN.py) automatically downloads the data set and provides the model frameworks as well as TODO sections that should be filled in.
For this code, the following libraries are required:
Jittor Conditional GAN (CGAN)
A Jittor implementation of a Conditional GAN (CGAN)
This code is based on a framework obtained from www.educoder.net/competitions/index/Jittor-6 for 第三届计图人工智能挑战赛. It is the warmup question “计图挑战热身赛”. The goal is to train a CGAN with the MNIST dataset, mapping noise and adding labels to number images from MNIST, that can then generate an image displaying the given input of numbers. The provided code framework (CGAN.py) automatically downloads the data set and provides the model frameworks as well as TODO sections that should be filled in.
For this code, the following libraries are required:
The installation instructions for Jittor can be found at https://cg.cs.tsinghua.edu.cn/jittor/download/
Or the libraries can be installed with
pip install -r requirements.txtTo run the code, the following command can be used
python CGAN.py