POSTECH-CVLab/PyTorch-StudioGAN

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

3487 stars · 344 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

PyTorch-StudioGAN is a PyTorch library providing unified implementations of representative GAN architectures (BigGAN, StyleGAN2/3, etc.) for conditional and unconditional image generation. It includes a large benchmark with pre-trained checkpoints, evaluation metrics like FID and IS, and a YAML-based configuration system for mix-and-matching GAN components.

Use cases

• train a StyleGAN2 or BigGAN model on CIFAR10 or ImageNet
• compare FID scores of different GAN architectures under identical settings
• evaluate a generative image model with clean-FID or PRDC metrics
• research new GAN losses or regularization techniques
• download pre-trained GAN checkpoints for image synthesis
• benchmark GANs against diffusion and autoregressive generative models

When to choose

• you need reproducible, unified implementations of many GAN variants for research comparison
• you want a comprehensive benchmark with pre-trained models and evaluation backbones
• you are doing academic research on generative image synthesis

When to avoid

• you need production image generation rather than research experimentation
• you want diffusion models only, as GANs are the primary focus
• you need a simple high-level API with minimal configuration

Facets

library · maturity stable

machine-learning deep-learning image-processing benchmarking deep-learning machine-learning image-processing artificial-intelligence python cross-platform gan pytorch image-synthesis generative-models stylegan biggan fid-evaluation research gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="POSTECH-CVLab/PyTorch-StudioGAN")