DCGAN tutorial here.
Real images | Fake images (epoch 0) | Fake images (epoch 10) | Fake images (epoch 20) |
---|---|---|---|
main.py
- Main file to run training of generator and discriminator, and generate fake images.models.py
- Utility. Model implementation (generator and discriminator).train_utils.py
- Utility. Functions to train models.tutorial.py
- Raw copy of the tutorial in PyTorch documentation.
python main.py [-b BATCH_SIZE] [-e EPOCHS] [-p PATIENCE] [-r LR] [-d DEVICE] [-s SEED] [--fraction FRACTION]
device
:"cpu"
/"cuda"
/"mps"
fraction
: How much of the data (CelebA) used to train. Change this to small values (e.g.,0.05
or0.2
) for a quick code check.