Skip to content

Unofficial implementation of Alias-Free Generative Adversarial Networks. (https://arxiv.org/abs/2106.12423) in PyTorch

License

Notifications You must be signed in to change notification settings

rosinality/alias-free-gan-pytorch

Repository files navigation

alias-free-gan-pytorch

Unofficial implementation of Alias-Free Generative Adversarial Networks. (https://arxiv.org/abs/2106.12423) This implementation contains a lot of my guesses, so I think there are many differences to the official implementations

Usage

First create lmdb datasets:

python prepare_data.py --out LMDB_PATH --n_worker N_WORKER --size SIZE1,SIZE2,SIZE3,... DATASET_PATH

This will convert images to jpeg and pre-resizes it. This implementation does not use progressive growing, but you can create multiple resolution datasets using size arguments with comma separated lists, for the cases that you want to try another resolutions later.

Then you can train model in distributed settings

python train.py --n_gpu N_GPU --conf config/config-t.jsonnet training.batch=BATCH_SIZE path=LMDB_PATH

train.py supports Weights & Biases logging. If you want to use it, add wandb=true arguments to the script.

Sample

Latent translation sample 1 Latent translation sample 2 Latent translation sample 3 Latent translation sample 4 Latent translation sample 5 Latent translation sample 6

About

Unofficial implementation of Alias-Free Generative Adversarial Networks. (https://arxiv.org/abs/2106.12423) in PyTorch

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published