Skip to content

jiamei-SHEN/GANbyPytorch

Repository files navigation

GANbyPytorch

An implementation of GAN using Pytorch on MNIST.

sample_from_testgan.png

中文说明

test_gan.py 大体是根据 B站课程 自己写的,有一些小小的改动。

gan_from_github.pygithub的一种实现,更新了一些老旧的用法。

output_images_bs_32_latentdim_64output_images_bs_100_latentdim_100 是 test_gan.py 中模型不使用 dropout 以及使用 ReLu(而不是LeakyReLU),在 batchsize 为 100(32), latent_dim 为 100(64)下的结果,效果不怎么样,但我不想删掉。

samples 下的 sample_from_testgan.png 和 sample_from_githubgan.png 分别是 test_gan.pygan_from_github.py 的输出结果。

English README

test_gan.py is written following Bilibili Video with a little modification.

gan_from_github.py is copied from an implementation from github, and I update some deprecated places.

output_images_bs_32_latentdim_64 and output_images_bs_100_latentdim_100 are results when the models in test_gan.py do not use dropout and ReLu(instead of LeakyReLU), with batchsize 100(32), latent_dim 100(64). These are bad results, but I don't want to delete them.

sample_from_testgan.png and sample_from_githubgan.png under samples are the output results of test_gan.py and gan_from_github.py respectively.

About

An implementation of GAN using Pytorch on MNIST.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages