CGAN_jittor

Brief Introduction：

This is a Jittor implementation of Conditional GAN (CGAN). I chose Jittor here because by using Jittor, users can build deep learning models conveniently. In addition, Jittor also has the following advantages: firstly, it is highly customizable and easy to use; secondly, the separation of coding and optimization provides users an opportunity to focus on using the front-end interface for coding, while the code is automatically optimized by the back-end; thirdly, all content is compiled immediately, so that users can change the source code at any time.

Default

--n_epochs = 100;
--batch_size = 64;
--lr = 0.0002;
--b1 = 0.5;
--b2 = 0.999;
--n_cpu = 8;
--latent_dim = 100;
--n_classes = 10;
--img_size = 32;
--channels = 1;
--sample_interval = 1000;

How to run the code:

Step 1:

Download Jittor from this website: https://cg.cs.tsinghua.edu.cn/jittor/download/

Step 2:

Run the following command in the terminal: python CGAN.py

Credit:

The structure of this code is provided by Tsinghua University Computer Vision Lesson Team. This is our homework, named “PA3: Conditional GAN”.