CGAN_jittor

A Jittor implementation of Conditional GAN (CGAN).

Prerequisite

You should install Jittor before running this program. See: cg.cs.tsinghua.edu.cn/jittor/download/

Usage

python CGAN.py [arguments]

Arguments

• --n_epochs: Number of epochs of training.
• --batch_size: Size of the batches.
• --lr: Adam: learning rate.
• --b1: Adam: decay of 1st order momentum of gradient.
• --b2: Adam: decay of 2nd order momentum of gradient.
• --n_cpu: Number of cpu threads to use during batch generation.
• --latent_dim: Dimensionality of the latent space.
• --n_classes: Number of classes for dataset.
• --img_size: Size of each image dimension.
• --sample_interval: Interval between image sampling.

Outputs

• result.png: Generated figures of number.
• <n>.png: Generated figures after <n> batches.
• generator_last.pkl: Pickle file of the latest generator.
• discriminator_last.pkl: Pickle file of the latest discriminator.