rosinality/stylegan2-pytorch

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2833 stars · 623 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A PyTorch implementation of StyleGAN 2, the image-quality-improved version of the StyleGAN generative adversarial network, including training, sample generation, latent projection, and weight conversion from official checkpoints. It also experimentally supports SWAGAN, a wavelet-driven variant.

Use cases

• train a StyleGAN2 model on my own image dataset
• generate realistic faces with a GAN in PyTorch
• convert official NVIDIA StyleGAN2 checkpoints to PyTorch
• project real images into GAN latent space
• discover interpretable latent directions in a GAN
• train SWAGAN for faster image generation

When to choose

• you want a faithful, widely-used PyTorch port of StyleGAN2 for research or training
• you need distributed multi-GPU training of a style-based GAN
• you want to reuse official TensorFlow checkpoints in PyTorch

When to avoid

• you need the latest GAN architectures or active maintenance and new features
• you want a no-code or GUI image generation tool
• your environment lacks CUDA GPUs, since training is GPU-oriented

Facets

library · maturity maintenance

machine-learning deep-learning image-processing deep-learning machine-learning image-processing python stylegan2 gan pytorch generative-models image-synthesis swagan gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="rosinality/stylegan2-pytorch")