transGAN

TransGAN: Two Pure Transformers Can Make One Strong GAN, and That Can Scale Up, arxiv

PaddlePaddle training/validation code and pretrained models for TransGAN.

The official pytorch implementation is here.

This implementation is developed by PaddleViT.

TransGAN Model Overview

Models Zoo

Model	FID	Image Size	Link
transgan_cifar10	9.31	32	google/baidu(9vle)

Notebooks

We provide a few notebooks in aistudio to help you get started:

*(coming soon)*

Requirements

• Python>=3.6
• yaml>=0.2.5
• PaddlePaddle>=2.1.0
• yacs>=0.1.8

Usage

To use the model with pretrained weights, download the .pdparam weight file and change related file paths in the following python scripts. The model config files are located in ./configs/.

For example, assume the downloaded weight file is stored in ./transgan_cifar10.pdparams, to use the transgan_cifar10 model in python:

from config import get_config
from models.ViT_custom import Generator
# config files in ./configs/
config = get_config('./configs/transgan_cifar10.yaml')
# build model
model = Generator(config)
# load pretrained weights
model_state_dict = paddle.load('./transgan_cifar10.pdparams')
model.set_dict(model_state_dict)

Generate Sample Images

To generate sample images from pretrained models, download the pretrained weights, and run the following script using command line:

sh run_generate.sh

or

python generate.py \
  -cfg=./configs/transgan_cifar10.yaml \
  -num_out_images=16 \
  -out_folder=./images_cifar10 \
  -pretrained=/path/to/pretrained/model/transgan_cifar10  # .pdparams is NOT needed

The output images are stored in -out_folder path.

Evaluation

To evaluate TransGAN model performance on Cifar10 with a single GPU, run the following script using command line:

sh run_eval_cifar.sh

or

CUDA_VISIBLE_DEVICES=0 \
python main_single_gpu.py \
  -cfg=./configs/transgan_cifar10.yaml \
  -dataset=cifar10 \
  -batch_size=32 \
  -eval \
  -pretrained=/path/to/pretrained/model/transgan_cifar10.pdparams  # .pdparams is NOT needed

Run evaluation using multi-GPUs:

sh run_eval_multi.sh

or

CUDA_VISIBLE_DEVICES=0,1,2,3 \
python main_multi_gpu.py \
    -cfg=./configs/transgan_cifar10.yaml \
    -dataset=cifar10 \
    -batch_size=32 \
    -data_path=/path/to/dataset/imagenet/val \
    -eval \
    -pretrained=/path/to/pretrained/model/transgan_cifar10  # .pdparams is NOT needed

Training

To train the TransGAN model on Cifar10 with single GPU, run the following script using command line:

sh run_train.sh

or

CUDA_VISIBLE_DEVICES=0 \
python main_single_gpu.py \
  -cfg=./configs/transgan_cifar10.yaml \
  -dataset=cifar10 \
  -batch_size=32 \

Run evaluation using multi-GPUs:

sh run_train_multi.sh

or

CUDA_VISIBLE_DEVICES=0,1,2,3 \
python main_multi_gpu.py \
    -cfg=./configs/transgan_cifar10.yaml \
    -dataset=cifar10 \
    -batch_size=16 \
    -data_path=/path/to/dataset/imagenet/train

Visualization of Generated Images

Generated Images after Training

Generated Images from Cifar10 datasets

Generated Images during Training

(coming soon)

Reference

@article{jiang2021transgan,
  title={Transgan: Two transformers can make one strong gan},
  author={Jiang, Yifan and Chang, Shiyu and Wang, Zhangyang},
  journal={arXiv preprint arXiv:2102.07074},
  year={2021}
}