Deep Convolutional Generative Adversarial Networks

StevenJokes · August 30, 2020, 12:12am

https://d2l.ai/chapter_generative-adversarial-networks/dcgan.html

StevenJokes · August 30, 2020, 12:19am

now it is not released.
http://preview.d2l.ai/d2l-en/PR-1309/chapter_generative-adversarial-networks/dcgan.html

merge now!
http://preview.d2l.ai/d2l-en/master/chapter_generative-adversarial-networks/dcgan.html

yoderj · September 3, 2020, 2:47pm

@StevenJokes Looks like it just got merged. Congratulations! I think it will show on the main book in a few days, right?

StevenJokes · September 3, 2020, 3:01pm

I know it has been merged.
I don’t know when it will show
I think it needs to be optim.
Usually pytorch is quicker than mxnet.
Help me if you can.
And find some ways to delete
@yoderj
BTW, I’m translating it to tensorflow
stuck one night…

github.com

StevenJokes/d2l-en-read/blob/master/chapter-generative-adversarial-networks/gan-tf1.ipynb

{
 "metadata": {
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.8-final"
  },
  "orig_nbformat": 2,
  "kernelspec": {
   "name": "python37864bittf2conda1de394c9651a4d19aa25641a40b8d2d8",
   "display_name": "Python 3.7.8 64-bit ('tf2': conda)"
  }
 },

This file has been truncated. show original

yoderj · March 28, 2023, 5:09pm

Hi, @StevenJokes. It’s been a long time and I want to give you a BRIEF update.

I’ve used this gan.py as the opening exercise in my Deep Learning course. Students don’t program anything – they are predicting runtimes. But some of them take a look at the images generated by the GAN when they complete the lab, and we show some of the generated images on our school (MSOE)'s supercomputer’s dashboard.

I really appreciate you making this work. It’s a nice “realistic” deep learning network that takes some time to train and produces an interesting result.