-
# hello, sorry to bother you again
## command line
python train.py --dataroot data_path --name model_name --model pix2pix --which_model_netG unet_256 --which_direction BtoA --lambda_A 100 --dataset_…
-
Hi, I am confuse about **Discriminator Loss**, in your paper ,
> We use the LSGAN objective proposed by Mao et al. [38]. We
employ multi-scale discriminators proposed by Wang et al. [20] to guid…
-
I get the idea of letting models setting the defaults that are specific to them; like the cyclegan using lsgan by default or the pix2pix using vanilla gan by default; but is the modify_commandline_opt…
-
After sucessufully compiled `./install.sh` I tried to reproduce`Generate 3D shapes, 2.5D sketches, and images` part by runing `bash ./scripts/figures.sh 0 car df` and got an error
```bash
error in…
-
https://github.com/eriklindernoren/Keras-GAN/blob/b817bb68993e21cdcf4581798b371b29dd0d90f2/cyclegan/cyclegan.py#L140-L159
I was wondering if there should be a sigmoid activation on the validity con…
Ouwen updated
5 years ago
-
The current code of [LSGAN](https://github.com/akanimax/pro_gan_pytorch/blob/master/pro_gan_pytorch/Losses.py#L175) loss does following:
```
0.5 * ((th.mean(self.dis(fake_samps, height, alpha)) - …
-
I modified the mnist.conf file and I was trying to run the model using the following command:
python train_network.py --config=configs/mnist.conf
But it is showing some argument error like this -
…
-
Hello,
Thanks for your work, it's fascinating. But I have a problem when training CVAE-GAN.
When does the model converge?g_acc close to 1 and d_acc close to 0?
And in mnist, I found that g_acc a…
-
Hello, thanks for providing the code. I am really interested in the work. I want to make the line generated by the model thinner. So I trained it as you inform. And I changed the aug_folder from width…
-
I used unet_256 to train pix2pix model.
I also applied 'scale_width_and_crop', 'load_size=1024' and 'crop_size=360' configuration.
Then there was a issue:
File "/home/hongjin/PycharmProjects/Cycl…