WGAN-GP

Title

Improved Training of Wasserstein GANs

Abstract

Generative Adversarial Networks (GANs) are powerful generative models, but suffer from training instability. The recently proposed Wasserstein GAN (WGAN) makes progress toward stable training of GANs, but sometimes can still generate only low-quality samples or fail to converge. We find that these problems are often due to the use of weight clipping in WGAN to enforce a Lipschitz constraint on the critic, which can lead to undesired behavior. We propose an alternative to clipping weights: penalize the norm of gradient of the critic with respect to its input. Our proposed method performs better than standard WGAN and enables stable training of a wide variety of GAN architectures with almost no hyperparameter tuning, including 101-layer ResNets and language models over discrete data. We also achieve high quality generations on CIFAR-10 and LSUN bedrooms.

Train

$ python main.py --mode train \
                 --scope [scope name] \
                 --name_data [data name] \
                 --dir_data [data directory] \
                 --dir_log [log directory] \
                 --dir_checkpoint [checkpoint directory]
                 --gpu_ids [gpu id; '-1': no gpu, '0, 1, ..., N-1': gpus]

---

$ python main.py --mode train \
                 --scope wgan-gp \
                 --name_data celeba \
                 --dir_data ./datasets \
                 --dir_log ./log \
                 --dir_checkpoint ./checkpoint
                 --gpu_ids 0

• Set [scope name] uniquely.
• Hyperparameters were written to arg.txt under the [log directory].
• To understand hierarchy of directories based on their arguments, see directories structure below.

Test

$ python main.py --mode test \
                 --scope [scope name] \
                 --name_data [data name] \
                 --dir_data [data directory] \
                 --dir_log [log directory] \
                 --dir_checkpoint [checkpoint directory] \
                 --dir_result [result directory]
                 --gpu_ids [gpu id; '-1': no gpu, '0, 1, ..., N-1': gpus]

---

$ python main.py --mode test \
                 --scope wgan-gp \
                 --name_data celeba \
                 --dir_data ./datasets \
                 --dir_log ./log \
                 --dir_checkpoint ./checkpoints \
                 --dir_result ./results
                 --gpu_ids 0

• To test using trained network, set [scope name] defined in the train phase.
• Generated images are saved in the images subfolder along with [result directory] folder.
• index.html is also generated to display the generated images.

Tensorboard

$ tensorboard --logdir [log directory]/[scope name]/[data name] \
              --port [(optional) 4 digit port number]

---

$ tensorboard --logdir ./log/wgan-gp/celeba \
              --port 6006

After the above comment executes, go http://localhost:6006

• You can change [(optional) 4 digit port number].
• Default 4 digit port number is 6006.

Results

• The results were generated by a network trained with celeba dataset during 10 epochs.
• After the Test phase runs, execute display_result.py to display the figure.

Directories structure

pytorch-WGAN-GP
+---[dir_checkpoint]
|   \---[scope]
|       \---[name_data]
|           +---model_epoch00000.pth
|           |   ...
|           \---model_epoch12345.pth
+---[dir_data]
|   \---[name_data]
|       +---000000.png
|       |   ...
|       \---12345.png
+---[dir_log]
|   \---[scope]
|       \---[name_data]
|           +---arg.txt
|           \---events.out.tfevents
\---[dir_result]
    \---[scope]
        \---[name_data]
            +---images
            |   +---00000-output.png
            |   |   ...
            |   +---12345-output.png
            \---index.html

---

pytorch-WGAN-GP
+---checkpoints
|   \---wgan-gp
|       \---celeba
|           +---model_epoch00001.pth
|           |   ...
|           \---model_epoch0010.pth
+---datasets
|   \---celeba
|       +---000001.jpg
|       |   ...
|       \---202599.jpg
+---log
|   \---wgan-gp
|       \---celeba
|           +---arg.txt
|           \---events.out.tfevents
\---results
    \---wgan-gp
        \---celeba
            +---images
            |   +---0000-output.png
            |   |   ...
            |   +---0127-output.png
            \---index.html

• Above directory is created by setting arguments when main.py is executed.