FlairGAN

マディさんとケビンさんの implementation of StyleGAN.

Instructions

1. Create a virtual environment in the root directory using the command:

   python3 -m venv .venv

2. Activate the virtual environment using the command:

   source .venv/bin/activate

3. Install the required dependencies using one of the following commands:

   pip3 install -r requirements_macos.txt
   pip3 install -r requirements_linux.txt

Cleanliness

Use black and isort to keep the code clean! They are installed as part of the development dependencies.

Training

We experimented with training on various platforms and found that the training speed varied widely between platforms. The following numbers were obtained using the original FCGAN network architecture.

• On a 2017 MacBook Pro with a 3.1 GHz Dual-Core Intel Core i5 processor and 8 GB 2133 MHz LPDDR3 RAM, each epoch took around 89.21 s.
• With a AMD Ryzen 7 3700X 8-Core Processor and 32 GB 3200 MHz DDR4 RAM, each epoch took around 56.77 s.
• Using an NVIDIA 3060 Ti GPU, each epoch took just 15.55 s.

Inference

To see the inference results on Linux:

• Make sure you SSH-ed with the -YC options to enable GUI commands.
• Install imagemagick with sudo apt install imagemagick.
• This gives you a tool called display.
• Use display to display your image from the command line.

Utilities

There is currently a utility script info.py that informs you whether or not CUDA is enabled on your system.

❯ python info.py
CUDA:   Yes